Full text of "The Wilson bulletin"

See other formats

HARVARD UNIVERSITY

Library of the
Museum of
Comparative Zoology

The Wilson Bulletin

PUBLISHED BY THE WILSON ORNITHOLOGICAL SOCIETY

VOL. 108, NO. 1 MARCH 1996 PAGES 1-204

(ISSN 0043-5643)

Thk Wii^oN Oknithoi.ogical Society
Founded Decemheu 3, 1888

Named after ALEXANDER WILSON, the first American Ornithologist.
President-Keith L. Bildstein, Hawk Mountain Sanctuary, RR 2, Box 191, Kempton, Pennsylvania
19529-9449.

First Vice-President— Edward H. Burtt, Jr., Department of Biology, Ohio Wesleyan University,
Delaware, Ohio 43015.

Second Vice-President— John C. Kricher, Biology Department, Wheaton College, Norton, Mas-
sachusetts 02766.

Editor— Charles R. Blem, Department of Biology, Virginia Commonwealth University, Richmond,
Virginia 23284-2012.

Secretary— John A. Smallwood, Department of Biology, Montclair State University, Upper Mont-
clair, New Jersey 07043.

Treasurer— Doris J. Watt, Department of Biology, Saint Mary’s College, Notre Dame, Indiana
46556.

Elected Council Memhers— Donald F. Caccamise and Laurie J. Goodrich (terms expire 1996),
Carol A. Corhat and William E. Davis (terms expire 1997), and Margaret C. Brittingham and
Herbert T. Hendrickson (terms expire 1998).

Membership dues per calendar year are: Active, $21.00; Student, $15.00; Family, $25.00; Sus-
taining, $30.00; Life memberships $500 (payable in four installments).

The W11.SON BUEt.ETIN is sent to all members not in arrears for dues.

The Josseeyn Van Tyne Memoihai, Libkary

The Josselyn Van Tyne Memorial Library of the Wilson Ornithological Society, housed in the
University of Michigan Museum of Zoology, was established in concurrence with the University
of Michigan in 1930. Until 1947 the Library was maintained entirely by gifts and bequests of
books, reprints, and ornithological magazines from members and friends of the Society. Two mem-
bers have generously established a fund for the purchase ol new books; members and friends are
invited to maintain the fund by regular contribution, thus making available to all Society members
the more important new books on ornithology and related subjects. The fund will be administered
by the Library Committee, which will be happy to receive suggestions on the choice of new books
to be added to the Library. William A. Lunk, University Museums, University of Michigan, is
Chairman of the Committee. The Library currently receives 195 periodicals as gifts and in ex-
change for The Wilson Bidlelin. With the usual exception of rare books, any item in the Library
may be borrowed by members of the Society and will be sent prepaid (by the University of
Michigan) to any address in the United States, its possessions, or Canada. Return postage is paid
by the borrower. In(]uiries and requests by borrowers, as well as gifts of books, pamphlets, reprints,
and magazines, should be addressed to: The Josselyn Van Tyne Memorial Librarv', University of
Michigan Museum of Zoology, Ann Arbor, Michigan 48109. Contributions to the New Book Fund
should be sent to the Treasurer (small sums in stamps are acceptable).

'Die Wh.son Buei.etin
(ISSN 0043-5643)

HIE II Util ,1 .E I IN (ISSN (K)4.t-.'>64;5) is [luhlislu'rj (jiiaricriy in March. June. Septcnilx'r. ami Deccniln'r by the Wilson
Dniilhologiral Society. f!IO Ea.sl lOtli Street. Diwrence. K.S WtO'W-HfWT. The subscription price, both in thi' United States and
elsewhere, is $40.00 per year. Second-class postaj;e paid at I jiwrence, KS. I’OSTM AS'l'Elt: Send address changes to THE WllisON
nUl.I.EriN. I’.O. Box IB07. Ijiwrence. KS 66(H4-8H07.

Back issues or single I'opies are available for $12. IK' etich. Most back issues of tbe liuUelin an- available and may be oniered
from tbe Treasurer. Special prices will 1m- (pioted for rpiantity orders.

.■\ll articles and cnmmnnications for publications. b(H>ks and publications for revit-ws should be addresst-d to tin- Editor. Ex-
chang.-s should he addressed to The josselyn Van Tyne Memorial l.ibrarv. Museum of Zoologv. Ann Arbor. Michigan 48100.

Suhsi riptions. changes of adilrcss and claims for undelivered copies should be sent to the DSNA. I’.O. Box 1807. Ijiwrence.
KS W)0H-8807. I’hone: (01.4) IH.4-1221; FAX: (01.4) (H.4-1274.

@ This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).

MCZ

LIBRARY

MaR 1 5 1996

HARVARD

UNIVERSITY

Frontispiece. The Chiribiquete Emerald, Chlorostilhon olivuresi. Male (above right) perched
on branch of Tepiiicmthus savannarunr, female (below left) feeding from flowers of Deccigon-
ocarpus cornutiim, on a mesa of the Siena de Chiribiquete, southeastern Colombia.

THE WILSON BULLETIN

A QUARTERLY MAGAZINE OE ORNITHOLOGY
Published by the Wilson Ornithological Society

VoL. 108, No. I March 1996 Pages 1-204

Wilson Bull., 108(1), 1996, pp. 1-27

A NEW SPECIES OF EMERALD HUMMINGBIRD
(TROCHILIDAE, CHLOROSTILBON) FROM THE
SIERRA DE CHIRIBIQUETE, SOUTHEASTERN
COLOMBIA, WITH A REVIEW OF THE
C. MELLISUGUS COMPLEX

E Gary Stiles'

Abstract. — The Chiribiquete Emerald {Chlorostilbon olivaresi sp. nov.) is described
from the Sierra de Chiribiquete, an isolated range of table-top mountains rising from the
flat Amazonian lowlands of the Departments of Guaviare and Caqueta, SE Colombia. This
hummingbird is a common inhabitant of the edaphic scrub and adjacent forests of the middle
and upper levels of the Sierra, but evidently does not occur in the suiTOunding lowlands. In
its morphology, the new species shows closer affinities with C. gibsoni of the Magdalena
Valley than with the adjacent cis-Andean populations of C. mellisiigus, but it is much larger
than all related forms. The Chiribiquete Emerald probably originated through the dispersal
of gibsoni-lypc birds to the Sierra during a dry period of the Pliocene or early Pleistocene,
perhaps in conjunction with hybridization with the local form of inelli.sugus\ large body size
probably evolved subsequently in the population as a response to its peculiar, insular habitat.
Variation in the Chlorostilbon mellisiigus complex in NW South America is described and
analyzed, and I conclude that the various forms are best treated as comprising a single
superspecies; melanorhynchus (including pumilus) of western Colombia and western Ec-
uador is sufficiently distinct from the adjacent assimilis and gibsoni. as well as from the
eastern forms of C. mellisugus to deserve (allo)species rank, and I suggest for it the English
name of West Andean Emerald. I recommend recognition of the following allospecies (from
north to south): auriceps, forficatus, canivetii, assimilis, melanorhynchus. gib.soni, olivaresi,
and mellisugus. Received 16 Feb. 1995, accepted 10 June 1995.

Abstracto. — Se describe Chlorostilbon olivaresi sp. nov. de la Sierra de Chiribiquete,
una serie aislada de mesetas en la planicie amazonica de los Departamentos del Guaviare y
del Caqueta, SE Colombia. E.ste colibri es comtin en las sabanas casmofitas y los bosques
aledanas de la parte media y superior de la Sierra, pero evidentemente no ocurre en los
bo.sques basales circundantes. Por sus caracteres morfologicos, C. olivaresi probablemente
tiene mas afinidad con C. gibsoni del Valle del Magdalena, que con las formas cisandinas

' Institute de Ciencias Naturales, Universidad Nacional de Colombia, Apartado 7495, Bogota, Colombia.

THE WILSON BULLETIN • Vol. 108, No. I, March 1996

vecinas de C. mellisugiis, pero es un ave mucho mas grande que cualquier forma de estas
especies. La poblacion de C. olivaresi posiblemente origino a traves de la dispersion de
un(os) individuo(s) del tipo gibsoni a la Sierra durante un pen'odo seco del Plioceno o
Pleistoceno temprano, tal vez incluyendo hibridizacion con la forma local de mellisugus',
subsecuentemente, el tamano corporal grande evoluciono en esta poblacion en respuesta a
su habitat peculiar y aislado. Se describen y se analizan los patrones de variacion en el
complejo de Chlorostilbon mellisugus en el NO de Suramerica, y se concluye que melan-
orhynchus (incluyendo a pumilus) del O de Colombia y Ecuador es suficientemente distinta
de gibsoni y assimilis, las formas adyacentes, como para ser considerado como una
(alo)especie. Recomiendo que el complejo de C. mellisugus se considere como una sola
superespecie, constituida por las aloespecies (de norte a sur) auriceps, forficatus, canivetii,
assimilis, melanorhynchus, olivaresi, y mellisugus.

The small hummingbirds of the genus Chlorostilbon are widespread in
the Neotropics from central Mexico to northern Argentina and the West
Indies. The genus is quite uniform in coloration: males of all species have
flashing green underparts, while females are pale gray below, nearly al-
ways with a distinctive facial pattern and dusky malar auricular area bor-
dered above by a white postocular stripe. Several species are easily dis-
tinguished by the distinctive form and color of the rectrices of the males
(e.g., poortmanni, alice, stenura), very bronzy or coppery coloration {rus-
satus), or extensively red bills {aureoventris). However, a number of
mostly or entirely allopatric forms in which the males have blue, more
or less forked tails, have long been a source of taxonomic confusion.
These forms, which collectively might be called the "'mellisugus com-
plex”, cover most of the range of the genus from Mexico to Bolivia and
eastern Brazil. Variation among them involves bill color and, in males,
the presence or absence of a glittering crown, the depth of the tail fork,
the shape of the outer rectrices, and the color of the underparts and, in
females, the shape of the rectrices and the amount of gray in the outer
rectrices (cf Zimmer 1950). Geographic variation in these characters pre-
sents something of a mosaic pattern, with similar forms often separated
by others of rather different appearance, making the determination of
species limits controversial. The discovery of a new form in the C. mel-
lisugus complex, here described as a new (allo)species, makes it desirable
to review the patterns of geographic variation in this complex in north-
western South America, and to reevaluate the relationships among the
various members of the complex as a whole in the light of recent studies.

The avifauna of Colombia has received at least as much attention from
ornithologists, both native and foreign, as has that of any large South Amer-
ican country. The birds of Colombia have been comprehensively mono-
graphed no less than three times (Chapman 1917, Meyer de Schauensee
1948-1952, Hilty and Brown 1986). Nevertheless, many parts of Colombia

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

have been visited only briefly or sporadically by ornithologists, and a num-
ber of areas remain ornithologically unexplored. Until very recently, one
such area was the Sieira de Chiribiquete, a small, isolated mountain range
in the Departments of Guaviare and Caqueta. Because of its topographic
uniqueness and pristine character, this area had been set aside as the Parque
Nacional Natural Chiribiquete in September 1989, but aside from a visit
by botanist Richard E. Schultes in 1943-1944, the Sierra remained biolog-
ically unexplored until the present decade. The only previous ornithological
collections from this entire region of Colombia were made by H. Romero
for the Institute de Ciencias Naturales of the Universidad Nacional de
Colombia at Araracuara, over 100 km S of the Sierra de Chiribiquete on
the Rfo Caqueta, in August-September 1977.

Between December 1990 and December 1992, three expeditions to the
Sierra de Chiribiquete were organized by the Agencia Espanola de Coop-
eracion Intemacional, the Instituto de Ciencias Naturales, and the Institute
de Recursos Naturales Renovables (INDERENA) of the Colombian govern-
ment. The first two expeditions, in December 1990 and August 1991, were
devoted to botanical, archaeological, and geological studies. During the first
expedition, the Colombian botanists P. Palacios and P. Eranco obtained a
specimen of an unusual hummingbird (the only bird specimen taken) that I
was unable to identify: it appeared to be the male of an undescribed form
of Chlorostilbon, notable for its very large size, but its rather rough prepa-
ration made precise comparisons with other material difficult.

I obtained additional material of this hummingbird during the third
expedition (18 November-2 December 1992). With J. L. Telleria and M.
Dfaz of the Universidad Complutense of Madrid, I observed and collected
birds in the vicinity of the expedition’s base camp in the northern part of
the Sierra at 0°56'N, 72°42'W at the site called “el Valle de los Menhires”
(Valley of the Monoliths) by Estrada and Euertes (1993). Seven specimens
(3 males, 4 females) of the Chlorostilbon were collected and eight others
measured and released; all specimens are housed in the collection of the
Instituto de Ciencias Naturales. On the basis of this sample, I here de-
scribe this hummingbird as

Chlorostilbon olivaresi, sp. nov.

CHIRIBIQUETE EMERALD

HOLOTYPE. — Adult male, no. 31266 of the ornithological collection of the In.stitiito de
Ciencias Naturales (original number FGS 2941), collected on 24 November 1992 in the
Valle de los Menhires, elev. 570m, Sierra de Chiribiquete, Depto. del Caqueta, Colombia
(0°56'N, 72°42'W) by E G. Stiles, J. L. Tellen'a, and M. Diaz.

PARATYPES. — One adult male (ICN 31252, orig. no. FGS 2927) and one sub-adult male
(ICN 31253, orig. no. FGS 2928) taken on 21 Nov. 1992 and three adult females (ICN

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

31244, 31245, and 31254, orig. nos. FGS 2915, 2916, 2929) and one subadult female (ICN
31243, orig. no. FGS 2914), taken on 19 and 24 Nov. 1992, all at the type locality.

DIAGNOSIS. — Clearly a member of the genus Chlorostilbon because of the entirely
steel-blue, forked tail of the males, and the distinct dusky face and cheeks and white post-
ocular stripe and uniform gray underparts of the females. Size significantly larger than any
other South American form of Chlorostilbon (exposed culmen always >18 mm, vs 17 mm
or less in all other forms); total culmen >21 mm vs <19.5 mm; wing chord >48 mm in
males, >47 mm in females, vs <47.5 mm and <46.5 mm respectively, except for the
isolated C. melUsugus duidae, which may reach 49.5 mm in both sexes; and weight usually
>3.5 g, vs <3 g). Differs from all forms of C. melUsugus in having the basal half of the
mandible red in males, or with a trace of red in females; in the dull, non-iridescent crown
(males) and extensively grey bases of the lateral rectrices (females). In these characters
resembles C. gibsoni from north and west of the Andes, but differs from all forms of gibsoni
in having the tail much more shallowly forked, with the lateral rectrices more truncate
(males) to rounded (females) at the tip, and the breast and throat much more strongly bluish
in color (males). In the latter feature is most like C. aureoventris of southeastern South
America, but always lacks red on the upper mandible (which is extensively red in the latter);
females of aureoventris also lack gray on the lateral rectrices.

ETYMOLOGY. — I take pleasure in naming this species for Fr. Antonio Olivares in honor
of his many pioneering contributions to Colombian ornithology and his indefatigable labor
in building the bird collection of the Institute de Ciencias Naturales. The English name
refers to the isolated mountain range which includes the type locality and evidently encom-
passes the entire distribution of the species.

DESCRIPTION OF HOLOTYPE.— (Color nomenclature follows Smithe 1975, 1981).
Crown, nape, back, and rump metallic green, near 162, Shamrock Green; a few feathers of
the nasal area of the anterior forehead more brilliant golden green (near 158, Chartreuse);
a small grayish-white postocular spot. Upper tail-coverts more bluish green (163, Emerald
green); tail dark steely blue, nearest 90, Blue-black. Facial area and sides of neck brilliant
Emerald Green, with Chartreuse reflections, passing abruptly to brilliant blue-green (between
164, Cyan, and 65, Turquoise Blue, depending on viewing angle) over the entire throat and
upper breast, this passing to brilliant green (near 62, Spectrum Green) on the lower breast
and belly; lower tail coverts more bluish, near Emerald Green; a small tuft of downy white
feathers on the thigh. Remiges blackish with faint bluish gloss (near 73, Indigo). Basal 3/4
or more of lower mandible red (between 13, Geranium Pink, and 10, Ruby); rest of bill,
legs, and feet black. Exposed culmen 19.5 mm, total culmen 21.8 mm, wing chord 49.6
mm, tail length 25.6 mm, tarsus 4.4 mm, weight 3.4 g. Adult male, left testis 2.4 X 2.3
mm, no fat; tiny diptera in stomach.

DESCRIPTION OF ADULT FEMALE (based on ICN 31244). — Upperparts somewhat
more bronzy-green than in the adult male, nearer 60 (Parrot Green), somewhat duller on the
crown; the more worn leathers of the back with more bluish tips (near Emerald Green),
producing a slightly scaly effect; longest upper tail coverts and most of central rectrices
more bluish (between Emerald Green and Cyan but darker); the second and third pairs of
rectrices similar, but shading to dark blue (near 173, Indigo Blue) at the tips. The outer two
rectrices are extensively pale gray basally (near 86, Light Neutral Gray), with the medial
portion Indigo Blue; the outermost rectrix is tipped broadly, the fourth and third progres-
sively more nairowly, with Light Neutral Gray. The malar and auricular areas are dark sooty
gray, tinged with dull bionze; a white stripe extends trom the eye back over the auriculars.
The underparts are pale gray, slightly tinged brownish (near 79, Glaucous, but paler), av-
eraging palest on the thioat and darkest on the upper breast. The basal 1/4 of the lower
mandible is tinged with dark red, the rest of the bill and feet are black.

Sliles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

PLUMAGE VARIATION IN THE TYPE SERIES.— Variation among the adults of the
type series is slight in both sexes, reflecting chiefly feather wear (the degree of scaliness
in the dorsal plumage, the degree to which the crown is dull and soiled in females); one
female (ICN 31245) has the tip of the first rectrix tinged with blue. The subadult male (ICN
31253) is approaching adult plumage, with the brilliant feathers below colored as in the
adults but more scattered over a dull, dusky green Juvenal plumage; traces of a dusky mask
and whitish postocular and malar stripes remain; the tail is less forked than in the adult
males, and the outer rectrix is very narrowly tipped with dull gray. The subadult female
(ICN 31243) is a more uniform, bronzy green dorsally with no bluish tinge to the upper tail
coverts or central rectrices, and a darker, duller gray below.

ADDITIONAL SPECIMENS EXAMINED.— Between May and August 1993, Diego Silva
and Tomas Walschburger of the Eundacion Puerto Rastrojo, conducted studies of birds in the
area of the Rio Mesay, just south of the border of the National Park (0°4'N, 72°26'W), some
85 km SSE of our study area. They observed and collected birds along a transect from the
river (ca 230 m elevation) to the top of a low mesa (ca 360 m), an isolated southern outlier
of the main Sierra. Four specimens of C. olivaresi were taken in the scrub atop the mesa in
July 1993 and were available for examination. These specimens agree perfectly in measure-
ments and coloration with the type series, allowing for differences in plumage wear: in par-
ticular, the golden-green reflections of the facial area and sides of the neck of the males are
stronger, increasing the contrast with the bluish green of the throat. The adult male specimen
from the 1990 expedition does not differ from the males of the type series. Thus, the characters
of C. olivaresi appear to be uniform over most or all of the Sierra de Chiribiquete.

THE SIERRA DE CHIRIBIQUETE AND THE
ECOLOGY OF C. OLIVARESI

The Sierra de Chiribiquete consists of a series of sandstone mesas and
buttes some 125 km long and 30 km wide, extending in an arc that curves
from NW-SE in the north to NE-SW in the south, between the latitudes of
1°20'N and 0°20'N, centered along the line of 73°W longitude (Eig. 1). The
mesas rise abruptly from the surrounding flat lowlands to heights of 800-
900 m in the north and 600 m in the south, often presenting several levels
or terraces separated by vertical cliffs. Many of the larger mesas are riven
by spectacular chasms or cracks; the flat upper surfaces are drained by
streams flowing through vertical cracks that emerge as waterfalls at the edges
of the mesas. The thin, sandy soil of the table-tops supports a scrubby veg-
etation interspersed with areas of naked rock: the stature of the vegetation
reflects the depth of the soil (or the absence thereof) at any given point.
Eurther details of the topography, vegetation, and geology of the Sieira de
Chiribiquete are given by Estrada and Euertes (1993). Extending south of
the main part of the Sierra are a series of progressively lower mesas (ca
300-350 m) which reach the Rfo Caqueta at Araracuara (ca 0°30'S).

The base camp of our expedition was situated on the flat middle level
of a large (ca 3 km long) mesa at an elevation of 570 m. To the north
and east, a line of cliffs rises abruptly to the top level of the mesa (ca
700 m); to the west, the mesa is bounded by a steep-sided canyon some

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Fig. 1. Left: location of the Sierra de Chiribiquete (inset) in relation to the major moun-
tain ranges (stippled) of Colombia, and the major rivers of SE Colombia and adjacent areas
(Roman numerals). The rivers are I. Orinoco; II. Guaviare; III. Vaupes; IV. Apaporis; V.
Caqueta; VI. Putumayo; and VII. Amazon. Other localities indicated are a. Magdalena Valley
and b. Sierra de la Macarena. Right: The Sierra de Chiribiquete and adjacent areas showing
major rivers and the collecting localities for Chloro.stilbon olivare.si. The 300, 400, 500, and
600 m contour lines are indicated. Localities: 1. Type locality (Valle de los Menhires); 2.
Collecting site of December 1990 specimen (Valle de las Abejas); 3. Rfo Mesay site. Re-
drawn from Estrada and Fuertes 1993, in part.

Fig. 2. Landforms and vegetation at the type locality. A. General topography, showing
the butte and our campsite, with adjacent stunted forest and surrounding Bonnetia scrub of
the middle level of the mesa; taller forest at the base of the fringing cliffs in foreground
and other me.sas just visible in the background. B. Aspect of Bonnetia scrub, the most
important habitat of C. olivare.si. Note areas of naked rock and patches of low vegetation
(mainly Navia garcia-harrigae) in foreground, the stiff, coriaceou.s-succulent leaves of many
shrubs (Grajfenriedia sp. in foreground, Cln.sia chirihiqueten.si.s at center. The low shrub L
of center is Decagonocarpti.s cornutu.s, the most important nectar source of C. oHvaresv,
taller shrubbery in background (with large white flowers) is Bonnetia niartiana.

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

THE WILSON BULLETIN • Vol. 108, No. 1. March 1996

100-150 m deep that slopes steeply down towards the surrounding low-
lands. The base camp was located some 100 m SE of the base of an
isolated butte that rises precipitously to a height of ca 75 m.

The vegetation around the base camp consisted of open scrub domi-
nated by the shrub Bonnetia martiana (Theaceae); patches of dense scrub,
in which the taller shrubs attained a height of 2-3 m, alternated with
patches of low, savannalike vegetation and areas of naked rock (Fig. 2).
The principal shrubs of the area had thick, coriaceous leaves and twigs
suggesting adaptation to resist drought, doubtless reflecting the shallow,
sandy soil with low nutrients and minimal water retention capacity. The
most common shrub species included, besides Bonnetia, Clusia chiribi-
qiietensis (Guttiferae), Tepuianthus savannensis (Tepuianthaceae), Graf-
fenriedia sp. (Melastomataceae), and Decagonocarpus cornutus (Ruta-
ceae). Between the shrubs, and in slight depressions where a thin layer
of sandy soil accumulated, occurred a low herbaceous vegetation domi-
nated by Xyridaceae, Eriocaulaceae, Burmanniaceae, Cyperaceae, and
Vellozia phantasmagorica (Velloziaceae) with only occasional grasses
(Gramineae); where drainage was impeded, pools formed after every rain
in which Utricnlaria spp. (Lentibulariaceae) were abundant. Areas of bare
rock were colonized by a terrestrial bromeliad, Navia garcia-barrigae. A
low, dense, tangled forest (canopy height 3-5 m) dominated by Clusia
spp. and Licania sp. grew around the base of the butte.

The level of the mesa sloped very gently towards the fringing cliffs to
the north and east, at the bases of which grew a much taller forest (canopy
height 20-25 m) with numerous palms and a relatively well-developed
understory; in the canyon bottom grew a forest of similar canopy height
but with few palms and notably low tree species diversity and a few large-
leaved monocots (Heliconia, Calathea, Costus, Phenakospennum) in the
understory. The northern part of the mesa was drained by a stream along
which grew an extremely dense, tall (canopy ca 5 m) stand of Bonnetia
whose tangled aspect suggested that of a young mangrove swamp. A
general inventory of the avifauna and its biogeographical affinities (Stiles
et al. 1995), and a detailed analysis of mist-net captures in relation to
vegetation characteristics (Diaz et al. 1996) will appear elsewhere.

In the study area, Chlorostilbon olivaresi was fairly common in the
open Bonnetia scrub of the mesa, where 12 of the 15 individuals were
captured; in fact, it was the bird most frequently captured in mist-nets in
the open areas of the mesa. In this habitat, both sexes were observed
visiting only the red-orange flowers of Decagonocarpus cornutus (see
Frontispiece) and were frequently noted gleaning small arthropods from
the foliage and flowers of Bonnetia and other shrubs and flycatching at
breaks in the vegetation, especially in late afternoon. Pollen samples taken

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

from the beaks of eight mist-netted birds with transparent Scotch tape
contained only pollen of Decagonocarpus, usually <10 grains (5 cases),
three grains of pollen of Decagonocarpus and one of Bonnetia (one biid),
or no pollen at all (2 birds). The visit to Bonnetia was almost certainly
to obtain insects such as thrips which were often present on the large,
open, and fragrant camellialike flowers of this species (which do not
produce nectar). Stomach contents of collected individuals contained tiny
insects, mainly flies, but a few thrips and/or microhymenopterans were
present in at least two stomachs. The only other hummingbirds to occur
regularly in the Bonnetia scrub were the Versicolored Emeiald {Amazilia
versicolor), which also visited Decagonocarpus but was much scarcer,
and the Black-throated Mango (Anthracothorax nigricollis), which ap-
peared to be more insectivorous and was seen to visit only the flowers
of a small tree of the Bombacaceae in the low forest at the base of the
butte. We never saw any interactions between any of these species.

The only other habitat in which we regularly recorded C. olivaresi was
the forest at the base of the fringing cliffs at the north end of the mesa.
Here, small numbers occurred in the understory and at gaps; three birds
were captured, and on several occasions a female was observed visiting
the flowers of a small understory tree of the Violaceae. In several days
of observation and netting, we never encountered C. olivaresi in the forest
of the canyon bottom. In visits to other areas at the base of the Sierra,
including mist-netting, other members of the expedition (A. Repizzo, B.
Ortiz) never encountered C. olivaresi in the forests at lower elevations.
During their work at the Rfo Mesay site, Silva and Walschburger only
recorded C. olivaresi in the Bonnetia scrub on the top of the mesa or in
the adjacent low forest (340-360 m), never lower or closer to the river.
The 4—6 months difference between the dates of our observations and
those of Silva and Walschburger would appear to preclude the possibility
of extensive seasonal movements (e.g., into the surrounding lowland for-
ests) by this hummingbird. Thus, C. olivaresi may be restricted to the
scrubby vegetation and adjacent forests on the mesas of the Siena, and
it is quite probably absent from the surrounding forested lowlands. It
evidently occurs widely in the Sierra de Chiribiquete, given the distance
between the Rfo Mesay site and the type locality; the 1990 specimen was
taken some 20 km south of the type locality, also in Bonnetia scrub at a
slightly lower elevation (ca 420 m). The southern limit of olivaresi re-
mains to be determined but is evidently somewhere between the Rfo Me-
say and the Rfo Caqueta, since it was not taken during intensive collecting
by H. Romero at Araracuara in similar Bonnetia scrub.

Of the specimens of C. olivaresi we collected, one adult male and two
females (as well as both of the subadults) had the gonads small and

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

undeveloped; one male and one female had moderately enlarged gonads.
Among the birds measured and released, one female had a developing
brood patch. The two subadult specimens were at least several months
old, as they had lost their bill corrugations. Taken together, these data
indicate that C. olivaresi was at the start of its breeding season during
our observations, although courtship displays were not observed, and we
found no active nests. None of the individuals captured were molting,
and most were in slightly to moderately worn plumage, consistent with
the inference that the breeding season was beginning (cf Stiles 1985b).
Although meteorological data for the Sierra de Chiribiquete do not exist,
extrapolation from the data for other sites in Colombian Amazonia sug-
gest that our expedition took place at the very beginning of the dry season
(Estrada and Fuertes 1993). Flowering of at least Decagonocarpus, evi-
dently the most important nectar source at this time for C. olivaresi, was
definitely increasing during our observations (many flower buds, few de-
veloping fruits). Fruiting of several species was also increasing or about
to begin in late November 1992, and a number of other bird species were
evidently just starting gonadal maturation, suggesting that the dry season
might include the main breeding period for a considerable segment of the
Chiribiquete avifauna (Stiles et al. 1995).

The four Rfo Mesay specimens, taken in July 1993, were for the most
part in very fresh plumage, and one female was in heavy molt; neither
of the two with gonad data were in breeding condition. If the seasonality
of the two sites is similar, this suggests that the breeding season of C.
olivaresi falls between late November or December and perhaps May.
Molt in the population extends from perhaps May through at least July,
as both males and one female ot the Rfo Mesay specimens had completed
molt by July, in hummingbirds it is not unusual for males to molt a month
or so ahead of females (cf Stiles 1985b).

Like other members of its genus, C. olivaresi is a rather quiet humming-
bird under most circumstances. TJie only vocalization heard was a shaip,
dry, scratchy cht , similar to the calls of other Chlorostilbon but somewhat
louder, and given by birds foraging at flowers. We never heard it sing.
Compared to others of the genus 1 have observed, C. olivaresi is less nervous
and flighty at flowers and shows much less of the incessant rapid flicking
or pumping of the tail while foraging (cf Stiles and Skutch 1989).

PATTERNS OF VARIATION IN THE C. MELLISUGUS COMPLEX IN
NORTHWESTERN SOUTH AMERICA

No fewer than 10 forms in this complex are recognized by most authors
frorn Colombia and adjacent areas of northwestern South America (see
able , Fig. 3). Because C. olivaresi appears to be a member of this

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

Table 1

Currently accepted Subspecies of the Chlorostilbon mellisugus Complex in
Northwestern south America, with their Distributions and Diagnostic characters'*

Subspecies

Distribution

Characters'

puniHus

W Colombia from Pacific slope
of W Andes E to W edge of
Magdalena Valley

Bill black. M: tail moderately
forked; glittering crown;
breast with little or no blue.

F: trace of grey in outer rec-
trices, extensive green in cen-
tral rectrices.

inelanorhynchus^

Extreme SW Colombia (Narino)
and W Ecuador

Like pumilus but averaging larg-
er.

gibsoni

Upper and middle Magdalena
Valley Colombia

Lower mandible largely red. M;
tail very deeply forked, outer
rectrices attenuate; breast
green; crown dull. F: outer
rectrices with extensive grey
bases, broad whitish tips.

chrysogaster

N lowlands of Colombia E to

Like gibsoni but M: tail even

Santa Marta; W of Lago de
Maracaibo, Venezuela-Colom-
bia

more deeply forked, fore-
crown glittering, throat and
breast tinged blue; F: grey of
outer rectrices darker, less ex-
tensive.

nitens^'

Arid N coast of extreme NE Co-
lombia and NW Venezuela

Like chrysogaster but tail of M
slightly less deeply forked.

caribaeus^

Most N coastal region of Vene-
zuela, S to Orinoco region
and Llanos of Colombia

Bill black. M; glittering crown;
breast with trace of blue; tail
shallowly forked. F; at most a
trace of dusky on lateral rec-
trices, otherwise blue.

napensis^

SE Colombia S of Llanos; E Ec-
uador; adjacent NE Peru

Like caribaeus but averaging
larger. M: tail very shallowly
forked, nearly truncate, breast
strongly tinged blue. F: exten-
sive green flecking on sides.

phaeopygos

E Peru to NE Bolivia, incl. adja-
cent Brazil

Like napensis but M; belly dull-
er, darker, less contrast with
blue of breast; F: much less
green flecking below. Also
averages larger, especially tail.

subfurcatus

S + SE Venezuela E to Guayana,
and adjacent NW Brazil

Like caribeus but M: tail less
forked, breast more bluish.
Averages slightly larger.

duidae

Mt. Duida, SE Venezuela

Like siibfuraliis but decidedly
larger; M: tail longer, more
forked; breast less bluish.

“ Characters and distributions from Meyer de Schauensec (1964, 1966); Meyer dc Schauensee and Phelps (1978); Zimmer
(1950). and Zimmer and Phelps (1952).

May not be distinct from pumihts.

‘ May not be distinct from clir\sof;asler.

Includes narttts from the Orinoco region.

'Often lumped with phaeopyf-os.

' M = male. F = female.

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

complex, I decided that to evaluate properly its status and affinities a
broader analysis of the patterns of variation in the complex over north-
western South America would be required.

Previous studies. — The first comprehensive study of the mellisugus
complex was that of Zimmer (1950) who concluded that the mosaic nature
of the patterns of geographic variation, plus the allopatric distributions of
most forms, justified considering all forms of the complex as subspecies
of a single, variable species mellisugus. However, several points of con-
fusion exist in his analysis. Evidently, Zimmer started by considering two
forms from northern South America, nitens and chrysogaster, as members
of different subspecies groups and later concluded that they might not be
separable at all, thus favoring lumping the “eastern” and “western”
groups.

In his initial study of the birds of Colombia, Meyer de Schauensee
(1948-1952) had considered all Colombian forms of the complex as sub-
species of gibsoni, but he later (1960, 1966) noted that two groups of
races could be distinguished on the basis of bill color: those with the
lower mandible largely red {gibsoni, chrysogaster, and nitens) and those
with all-black bills (all other forms). He noted that the red-billed forms
were also those with the most deeply forked tails and attenuated outer
rectrices in the males, and with the most gray in the rectrices of the
females. He noted apparent sympatry of a red-billed (gibsoni) and black-
billed ipumilus) form without evident intergradation at two localities on
the western edge of the Magdalena Valley in Colombia and concluded
that two species should be recognized in Colombia: the red-billed gibsoni
(including chrysogaster and nitens) and the black-billed mellisugus (in-
cluding pumilus, melanorhynchus, caribaeus, and phaeopygus {=napen-
sis).

Wetmore (1968) adopted Meyer de Schauensee’s character of bill color
as a criterion for distinguishing species in the complex when he separated
the black-billed assimilis of southwestern Costa Rica and W Panama from
the red-billed canivetii of farther north. The arrangement of Meyer de
Schauensee (1960, 1966) has been followed by recent authors for the
South American forms of the complex (Meyer de Schauensee and Phelps

Eig. 3. Tails of members of the C. melli.sugu.s complex from Colombia, drawn to the
same scale. Left: tails of adult males, showing form of rectrices; in all, rectrices are uniform
steely blue-black. Right: tails of adult females, showing form, pattern, and colors of the
rectrices. Solid; steely blue-black; heavy stipple: green; light stipple: grey; open: white. The
forms illustrated are: a. chrysogcLster, b. gibsoni', c. olivaresi', d. napensis', e. caribaeus-, f.
pumilus.

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

THE WILSON BULLETIN • Vol. 108, No. I, March 1996

1978, Hilly and Brown 1986, Sibley and Monroe 1990). However, be-
cause the characters of olivaresi, as detailed above, appear to be in part
a mosaic of features of gibsoni and mellisiigus as cuirently recognized,
this arrangement requires a detailed reevaluation. For this analysis, I con-
sulted specimens of the mellisiigus complex available in the following
museums (numbers of specimens in parentheses): the Institute de Ciencias
Naturales, Universidad Nacional de Colombia (69); the Universidad de
La Salle, Bogota (12); the Unidad de Investigacion “Federico Medem”
(UNIFEM), INDERENA (6), Bogota; and the Colegio de San Jose, Me-
dellin (8). Eor each specimen, I measured the exposed culmen, total cul-
men, wing chord, and tail length to the nearest 0.1 mm with dial calipers.
Eor males only, I measured the depth of the tail fork as the difference
between the lengths of the first and fifth rectrices. In addition, I took
detailed notes on color characters of each form and made detailed draw-
ings of the rectrices of representative individuals of both sexes of all
forms (Eig. 4). Eield work in various parts of Colombia has also given
me the opportunity to capture and measure an additional 18 individuals
of various forms of the mellisiigus complex. Eor these birds, I made the
preceding measurements except depth of tail fork, also with dial calipers,
and to the same level of accuracy; I also weighed all birds to the nearest
0.1 g with a 10 g Pesola spring balance. Because in previous studies I
had found that my field measurements did not differ from those taken in
the museum and that my measurements of birds in the field agree closely
with remeasurements of the same birds prepared as museum specimens
(Stiles 1985a, 1995), I have included both types of measurements in the
quantitative analyses below. In addition, M. Marin kindly measured 10
specimens of napensis and pheaopygus in the collection of the Museum
of Natural Science, Louisiana State Univ. Having worked with and mea-
sured birds previously with Marin, I am confident that his measurements
are comparable with mine and have incorporated them into the analyses.

In all, measurements from 1 1 1 specimens of the following forms were
used in the analyses: olivaresi, pumilus, gibsoni, chrysogaster, caribaeus,
and napensis (Table 2). Of the remaining forms in Table 1, none occurs
in areas adjacent to olivaresi and is likely to bear upon its status; I was
able to examine only two specimens of nitens, and none of melanorhyn-
clnis, subfurcatus, duidae, or phaeopygus. Eor the characters of these
forms (Table 1), 1 have relied upon the descriptions of Zimmer (1950)
and Zimmer and Phelps ( 1952). For the six forms mentioned above, mea-
surements were analyzed by one-way analysis of variance (ANOVA) for
each sex separately. Where a significant result was obtained in the ANO-
VA, Tukey a posteriori tests were performed to determine which forms
dilfered significantly with respect to the measurement in question (Zar

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

s purmliLS
Wh mtlanorhy'nchus
^ gibsoni

chryso^cLster

nitens
caribae^us
subfurcatws
A dmdoL
• oUvatzsi
napcnsis
'////, phaeopggos

Eig. 4. Distributions of the members of the C. mellisugus complex in northwestern South
America. Note that limits of some forms are imperfectly known, as are possible zones of
contact; especially east of the Andes, distributions are probably more continuous than shown,
but collecting localities are often widely scattered.

1988). For the measurement of depth of tail fork, I derived a measure of
“relative tail fork” by dividing the difference r5-rl by the tail length for
each specimen, to control for differences in absolute size between the
various forms.

The most striking result of the analyses of variance of the measure-
ments (Table 3) is the clear-cut separation of olivaresi (at a significance
level of P < 0.001) from all other forms of the Chlorostilhon mellisugus
complex, with respect to bill length (both exposed and total culmen), wing

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

•7

«3

-g

+ !

+ 1

iT)

r-;

»— «

V ^

> t

-- ,o
— ^

“v ^

2 5

'tt

■« ^

+ 1

d
+ 1

r--

-J

—

— 1

r— .

•o

::d

1—

acn
■S o

+ 1

+ i

+ 1

»-g

-J

'tt

»r)

•-J

3 c

+ 1

+ !

+ 1

c2 -

H a

1-^

•-H

—

1— M

IT)

Tj-

8.^

+ !

X 3

loi

■ — '

— H

•—

Sex

U-.

P-i

P-

U-

•-“v

Form

-5

-i:)

•S!

§ ti

Mean and standard deviation. Masses are in g; culmen, wing chord, and tail length are in

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

Table 3

Result of Analyses of Variance Comparing Measurements of Members of

ChLOROSTILBON MELUSUGUS COMPLEX FROM NORTHWESTERN SOUTH AMERICA

Measurement

Sex F

Result of Tukey tests”

-/ ///-

ExpKjsed culmen M 109.19 <0.001 PU CA CC CN GI NA OL

-///-

F 94.75 <0.001 PU CA GI NA CC CN OL

Total culmen

-///-

M 53.90 <0.001 PU CA CC CN NA GI OL

Wing chord

-///-

F 71.83 <0.001 PU NA CA GI CC CN OL

-II-

-h

-///-

M 48.81 <0.001 CC CA NA CN PU GI OL

-/-

-///-

F 65.53 <0.001 NA CA CC CN PU GI OL

Tail length

-/// //

M 53.08 <0.001 CA NA PU OL GI CC CN

-II-

F 17.57 <0.005 NA CA PU GI CC OL CN

Body mass

-///-

M 29.64 <0.001 PU CA NA GI OL

-///-

F 42.47 <0.001 NA CA PU GI OL

-///-

Relative tail fork M 103.89 <0.001 NA OL CA PU GI CN CC

“ Abbreviations: CA = carihaeus', CC = chrysogaster — N coast of Colombia; CN = chrysogaster — Norte de Santander,
NA = napensis\ GI = gihsonv, OL = olivaresr, PU = pumilus.

Forms not significantly different (P < .05) in a given measurement are connected by solid lines. Breaks in the lines
indicate significant differences, and the number of slashes indicates the degree of significance: one slash = P < 0.05; two
slashes = P < 0.01; three slashes = P < 0.001.

chord, and mass. The larger size of olivaresi is most dramatically shown
by the data on body mass, perhaps the best indicator of overall size; this
form is fully 35-40% heavier than all the other forms, which do not differ
among themselves. In fact, olivaresi is apparently the largest form in the

THE WILSON BULLETIN • Vol. 108, No. I, March 1996

entire genus Chlorostilbon, exceeding in size even the Cuban Emerald C.
ricordii (cf Ridgway 1911, Howell 1993). Regarding measurements of
the tail, the separation of olivaresi is more equivocal, chiefly because this
form, because of its overall large size, has a tail length more similar to
those of the gibsoni group with their relatively long, deeply forked tails,
but in its shallowly forked tail, olivaresi resembles closely various mem-
bers of the mellisugus group.

Although some of the remaining forms differed significantly from oth-
ers in one or another dimension, similar clean-cut separations between
groups were the exception. Males of the gibsoni group differed from those
of the mellisugus group significantly in tail length and highly significantly
in relative depth of tail fork; however, in most other dimensions the pat-
tern was less clear: often differences significant in one sex were not so
in the other, and the rank order of the forms often varied from one mea-
surement to another and/or between the sexes for a given measurement
(Tables 2, 3). This is undoubtedly at least in part a reflection of the mostly
small sample sizes (Table 2). One interesting result was the significant
difference in wing length between the northern and cis-Andean popula-
tions of chrysogaster, however, these populations were similar in other
dimensions. Aside from olivaresi, the most distinctive form, in terms of
measurements, was pumilus, with its unique combination of a very short
bill and a relatively long wing (Table 3; see below).

Patterns of plumage variation. — In eastern Colombia, specimens of
caribaeus showed no obvious variation between Arauca and Meta (in
particular, those from Arauca showed no tendency to approach chryso-
gaster). To the south, napensis differed in having a shorter, less forked
tail and more strongly bluish throat in the males, more lateral green fleck-
ing below and less pale tipping on the outer rectrices in the females, and
a somewhat longer bill in both sexes. I could detect no consistent differ-
ences in size or color between specimens of napensis from extreme south-
ern Meta, northwestern Caqueta, eastern Vaupes, or southern and eastern
Amazonas in Colombia, or (in size at least) eastern Ecuador or extreme
northeastern Peru. The largest male measured was from eastern Peru,
where the range of this form approaches that of the possibly indistin-
guishable phaeopygos. I have not seen specimens of phaeopygos from
Peru, so cannot comment on the validity of separating napensis', however,
female specimens of the latter I have seen do show the extensive green
flecking below used by Zimmer (1950) to distinguish this race from
phaeopygos and caribaeus. Although napensis and caribaeus differed sig-
nificantly in relative tail fork in this analysis (Table 3), it is worth noting
that this difference is apparently bridged by the adjacent subfurcatus and
duidae, to judge from the measurements of these forms given by Zimmer

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

and Phelps (1952). I should note in passing that my measurements are in
reasonably close agreement with those of Zimmer for those forms which
we both measured.

The range of olivaresi appears to be nearly completely nested within
that of napensis. Although I have seen no specimens of the latter from
localities adjacent to the Sierra de Chiribiquete, those from northwest,
south, and east of this range show no approach to the characters of oli-
varesi; in fact, napensis differs from olivaresi in most characters at least
as much as does any other cis-Andean form of the mellisugus complex.
This indicates that gene flow between olivaresi and napensis does not
occur; given the great difference in size, it is highly questionable whether
interbreeding could occur at all.

Within the gibsoni group, chrysogaster males have a glittering fore-
crown and usually, at least, a faint bluish tinge to the throat (more than
either pumilus or gibsoni). The tail is even more deeply forked than in
gibsoni; in females, the pattern of the tail is similar, although the gray
area at the bases of the outer rectrices tends to be darker and less exten-
sive. The difference in wing length between the populations of this form
east and west of the Serrania de Perija is interesting: at the very least, it
appears that the former shows no approach to the geographically adjacent
form of mellisugus, caribaeus. In color, pattern, and tail shape, the cis-
Andean population of chrysogaster also appears as different from cari-
baeus as does the trans-Andean population, suggesting that gene flow
between these populations is not occurring. The little information I have
on nitens suggests that this form is very similar to the trans-Andean chry-
sogaster. Although data are required on possible zones of contact between
nitens and caribaeus, the information currently available again favors
maintaining separate species status for the gibsoni group (including chry-
sogaster and nitens) and mellisugus.

The case of C. pumilus. — This form of western Colombia agrees with
members of the mellisugus group in having an all-black bill, in the glit-
tering crown of the males, and in the virtual absence of grey in the outer
rectrices of the females. However, the males have significantly longer,
more deeply forked tails and less bluish breasts than do any of the eastern
forms of this group. Females of pumilus differ from those of the eastern
races in their slightly forked to double-rounded tails, with extensive green
in most rectrices, and in the more extensive pale tips to the outer rectrices,
as well as in showing some dark gray at the bases of these rectrices. In
overall proportions, pumilus has a shorter bill and longer wing and tail
than do the eastern races, sex for sex.

In several of these characters, pumilus rather resembles the adjacent
gibsoni of the Magdalena Valley; however, it differs strongly from gibsoni

THE WILSON BULLETIN • Vol. 108. No. I, March 1996

in bill color, tail shape, bill length, crown color (males), and the color of
the outer rectrices (females). Moreover, these differences appear to be as
great in specimens from areas where the two forms approach each other
(western Tolima, western Antioquia), as in areas far from the possible
contact areas. Hence, the evidence presently available favors continuing
to separate pumilus and gibsoni at the species level.

Considering the next form to the north, assimilis, pumilus also shows
a number of clear-cut differences based upon the measurements, weights,
and descriptions of the latter in Wetmore (1968). In males, the crown of
pumilus is highly iridescent, while in assimilis it is plain; the gray of the
outer rectrices of the females is much more extensive in pumilus. In mea-
surements, pumilus has a much shorter bill, longer wing, and shorter tail
and also weighs considerably less in both sexes. The ranges of the two
forms are separated by a wide gap in eastern Panama (cf Wetmore 1968)
but from my observations of both forms in the field, I suspect that they
differ in ecology, with pumilus preferring significantly wetter areas than
does assimilis. I conclude from this analysis that pumilus is best consid-
ered an allospecies in the mellisugus superspecies rather than as a sub-
species of mellisugus (sensu stricto); this also eliminates the disjunct dis-
tribution of the latter (with pumilus separated by gibsoni from the other
forms of mellisugus).

Although I was unable to examine specimens of melanorhynchus, there
seems little reason to doubt that this form is conspecific with pumilus',
Zimmer (1950) noted that the two differ little, if at all, in coloration and
overlap in measurements and, in fact, might not be separable. The two
forms were supposed to differ in their preferred elevations, with melan-
orhynchus occurring mostly above 2000 m, and pumilus at lower eleva-
tions; however, this difference also tended to break down in the series of
specimens available to Zimmer (1950). If pumilus and melanorhynchus
are considered to comprise a separate allospecies in the mellisugus com-
plex, as I feel the evidence indicates, this species will have to be called
C. melanorhynchus (since melanorhynchus Gould 1860 has priority over
pumilus Gould 1872). Without having seen specimens of melanorhynchus
itself, I cannot comment on the advisability of synonymizing pumilus and
recommend further study.

On the basis of a phylogenetic species concept (Cracraft 1983, Mc-
Kitrick and Zink 1988), melanorhynchus (with pumilus) would also clear-
ly be entitled to species status; it possibly differentiated from other forms
of the mellisugus complex in the Choco humid forest refugium during
one of the dry epochs of the Pleistocene (cf Haffer 1974). I suggest the
English name West Andean Emerald for this form in view of its basically

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

Andean distribution, which lies to the west of all other members of the
mellisugus complex in South America.

Origin of the Chiribiquete Emerald. — C. olivaresi represents an anom-
aly among the cis-Andean populations of the mellisugus complex: its
characters are sharply discordant with the general trends of variation over
this broad area. Although large size is a character shared by another iso-
lated mountain endemic, duidae of Venezuela, the differences in color
and morphology of the latter relative to the forms occurring elsewhere in
southern and southeastern Venezuela, caribaeus and particularly subfur-
catus, are differences of degree only: there seems little reason to doubt
that duidae is but a derivative of one or the other (apparently neither has
been recorded from the adjacent lowlands), and there is little reason to
suspect that it is reproductively isolated from them (Zimmer and Phelps
1952). In fact, duidae was not considered by Mayr and Phelps (1967),
although it was recognized as distinct by Meyer de Schauensee and Phelps
(1978). The case of olivaresi is clearly not comparable to that of duidae:
it is certainly not an isolated derivative of napensis (or any other cis-
Andean form). The many points of resemblance to gibsoni (bill color, the
dull crown of the males, and the rectrix color of the females) suggest that
the origins of olivaresi might lie with that form, at least in part. At the
present time, the ranges of olivaresi and gibsoni are separated by at least
300 km of forested lowlands and foothills, which constitute unsuitable
habitat for both forms which are evidently adapted to dry or edaphically
scrubby habitats (and which are inhabited, at least at present, by napen-
sis). It is likely that the eastern foothills of the Andes in southeastern
Colombia represented a humid enclave or “forest refugium” through
Pliocene-Pleistocene times (Haffer 1974, 1985), such that the range of
gibsoni has probably never included or approached the Sierra de Chiri-
biquete. A more likely scenario is for one or a few individuals of the
ancestral population of gibsoni to have dispersed across the forested low-
lands to Chiribiquete, perhaps during a dry period in the Pliocene or early
Pleistocene. During such a period, the low passes south and west of the
Sierra de la Macarena might well have supported a more xeric, scrubby,
vegetation suitable for gibsoni, facilitating its arrival on the eastern slope
of the Andes. Consisting of sandstones of Permian age or earlier, the
Sierra de Chiribiquete antedates at least the final uplift of the Andes (cf
Gal vis, in Estrada and Fuertes 1993), and would have been available for
colonization by forms adapted to xeric or open habitats throughout the
latter part of the Cenozoic. The final population of Chlorostilbon to col-
onize the Sierra de Chiribiquete might even have been the result of hy-
bridization of the newly arrived gibsoni-iype birds with the form of mel-
lisugus already present in the adjoining areas. At least, this would provide

THE WILSON BULLETIN • VoL 108, No. I, March 1996

a tentative explanation for the meUisugus-\\](..& characters of olivaresi (very
shallowly forked tail and blue breast of the males). Once established, this
population of possibly hybrid origin might well have developed large size
in the process of adapting to its peculiar insular habitat. Large size is a
feature of some island populations (cf Grant 1965) including those of
Chlorostilbon (cf Ridgway 1911, Howell 1993), and the large size of
duidae may represent a result of the same processes, albeit at an earlier
and more incomplete stage. A possibly analogous case of presumptive
hybrid origin of a currently stable and well-differentiated form among the
white-eyes of Reunion Island has been discussed by Gill (1973). Genetic
studies in the mellisugus complex would assuredly shed light on the or-
igins of olivaresi, and might help to resolve several other questions re-
garding geographical variation and species limits in the complex. In any
case, by both biological and phylogenetic criteria, olivaresi appears as
entitled to separate (allo)species status as does any other form within the
entire mellisugus complex.

Among the avifauna of the Sierra de Chiribiquete, C. olivaresi is also
an anomaly. Virtually all of the species of the upper levels of the Sierra
whose subspecific allocations could be determined, belong to forms
whose distributions include the Llanos, the Orinoco region, and/or the
savannas of eastern Vaupes and adjacent Brazil, rather than the Amazo-
nian lowlands and Andean foothills to the west and south (Stiles et al.
1995). The lack of differentiation of these forms suggests a vicariance
pattern, probably a relict of more a continuous distribution of savanna or
scrub-adapted birds during the dry periods of the Pleistocene. The present
isolation of these birds from their relatives to the north and east probably
dates back no more than 15,000—20,000 years, much too short a time to
have permitted significant differentiation among birds of the Amazon
drainage (Capparella 1988).

Interestingly, one other species of Chiribiquete appears to have dis-
persed from the xeric upper Magdalena Valley rather than having its af-
finities with forms to the north and east. This species, Hemitriccus mar-
garitaceiventer, also appears to have differentiated in Chiribiquete from
its putative ancestral population in the Magdalena Valley, although to a
much lesser extent. This population clearly represents a distinct subspe-
cies of margaritaceiventer (Stiles, unpubl. data), probably of more recent
origin than C. olivaresi. In general terms, the Chiribiquete avifauna is
basically a relict of a formerly more continuous Orinoquian avifauna upon
which is superimposed a small number of forms that have dispersed from
similar habitats to the west, at different periods of the geological past (cf
Mayr and Phelps 1967). The most distinctive, and probably the oldest, of

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

these forms is undoubtedly the Chiribiquete Emerald, Chlorostilbon oli-
varesi.

AN OVERVIEW OF THE C. MELLISUGUS COMPLEX

The most recent treatment of this complex as a whole is that by Sibley
and Monroe (1990). These authors consider as allospecies of the melli-
sugus superspecies canivetii of Mexico and northern Central America,
assimilis of southern Costa Rica and western Panama, and mellisugus
(including pumilus and melanorhynchus) of northern South America,
while excluding gibsoni (including chrysogaster and nitens). This repre-
sents the opposite extreme from the broad treatment of Zimmer (1950)
who considered all these forms as subspecies of mellisugus. Especially in
view of the characters of olivaresi, which combines certain features of
both gibsoni and the mellisugus group, I would agree with Zimmer that
all of these forms are representatives of a common stock. In particular, I
see no justification for including the canivetii group while excluding gib-
soni from this complex: they share characters, such as the red lower man-
dible and deeply forked tail of the males, and extensive gray bases and
broad white tips to the lateral rectrices of females (Zimmer 1950). These
features could well represent parallel adaptations to more open, seasonal
habitats than are occupied by neighboring forms.

In general, I feel that too little attention has been paid to the ecology
of these hummingbirds — in particular, the range of elevations and humid-
ity conditions occupied by each form. Although all members of the group
prefer open, brushy habitats, so far as known, there are a number of
differences that could have bearing on their status. Eor instance, pumilus
appears to be primarily a bird of humid foothills, entering well into the
subtropical zone where it has been recorded to elevations of 2000 m or
more (Hilty and Brown 1986; pers. obs.); the same evidently is true of
melanorhynchus (Zimmer 1950). The preferred habitat of gibsoni appears
to be in hotter, drier areas; it occupies the floor of the middle and upper
Magdalena Valley, up to ca 2000 m or more in dry valleys, mainly on
the western side; chrysogaster and nitens are mainly or exclusively low-
land forms, with the latter occupying the driest areas along the northern
coast (Meyer de Schauensee 1948-1952). Also occurring primarily in the
lowlands east of the Andes are caribaeus (probably throughout the Lla-
nos) and napensis in more humid areas farther south, apparently largely
in areas of riverine scrub. Subfurcatus, by contrast, occurs mostly above
1000 m in the Gran Sabana and tepuis of eastern and southeastern Ven-
ezuela and adjacent Brazil and Guyana (Meyer de Schauensee and Phelps
1978); duidae is evidently isolated at similar elevations on Cenro Duida
(Zimmer and Phelps 1952). Different habitat preferences may help to

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

promote or reinforce reproductive isolation of different forms where they
meet, as perhaps is the case between pumilus and gihsoni on the western
side of the Magdalena Valley. This would likely be the case for napensis
and olivaresi, which appear restricted to the upper elevation Bonnetia
scrub of the Sierra de Chiribiquete. A detailed study of the displays of
these hummingbirds (cf Stiles 1983) might also help to clarify species
limits.

On present evidence, I recommend that the relationships of these forms
are best expressed by considering the mellisugus complex to comprise a
single {mellisugus) superspecies. In South America, the forms melano-
rhynchus (including pumilus), gibsoni (including chrysogaster and ni-
tens), olivaresi, and mellisugus (including caribaeus, napensis, phaeo-
pygos, subfurcatus, duidae, and one or more races in eastern South Amer-
ica beyond the scope of this study; see Zimmer 1950) should be consid-
ered allospecies within this superspecies.

In this connection, an interesting analysis of the northern forms of the
mellisugus complex has recently appeared (Howell 1993). Using argu-
ments similar to those employed here, Howell has presented strong evi-
dence for recognizing auriceps of western Mexico and forficatus of Coz-
umel Island, as distinct species. He also advocates recognizing salvini of
northern Central America as distinct from canivetii of southeastern Mex-
ico and extreme western Guatemala (considering the two to comprise a
superspecies). In the latter case, however, depth of tail fork (in both sexes)
appears to be the only reasonably clear-cut difference. This difference is
on the same order of magnitude as that between caribaeus and napensis
(in fact, the mean ratio of the relative tail forks of males of the latter two
forms is ca 2.19, whereas the corresponding mean ratio for canivetii and
salvini is only 1.30). The differences in patterns of the rectrices cited by
Howell between the two forms (cf his hg. 1) are of a magnitude that, in
my experience, could be subsumed by individual variation. In the absence
of a more detailed analysis of geographic variation within each form, 1
am hesitant to follow Howell (1993) in separating these forms at the
species level. For the present, I prefer to adopt a more conservative spe-
cies criterion and consider canivetii and osberti (the form of salvini in
question) as only subspecifically distinct, particularly as there does not
appear to be any significant ecological difference between them; this is
also more consistent with species criteria in the mellisugus complex as a
whole (contra Howell 1993). This problem clearly merits further study.
Also, in keeping with the mosaic pattern of variation in this complex as
discussed here, I consider auriceps and forficatus as allospecies of the
mellisugus superspecies, rather than as species apart.

I should note here that I had previously (Stiles and Skutch 1989) ques-

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

tioned the distinctness of assimilis from canivetii, based upon apparently
intermediate birds that I had observed and trapped (but in most cases not
collected) along the Pacific coast of southcentral Costa Rica. While wide-
spread hybridization and introgression does occur between Hoffmann’s
Woodpeckers Melanerpes hojfmanni of northwestern Costa Rica and Red-
crowned Woodpeckers (M. mbrucapillus) of the southern Pacific area
following deforestation (cf Stiles and Skutch 1989; various specimens), I
now feel that the evidence for the same in Chlorostilbon is still too weak
to justify lumping assimilis and canivetii and recommend further study
of this problem. For the present, it seems best to follow current practice
(A.O.U. 1983, Sibley and Monroe 1990) in considering these forms to
be allospecies; indeed, a limited zone of hybridization would not be in-
consistent with this interpretation.

In conclusion, I recommend that the ^‘'mellisugus complex be recog-
nized as a single superspecies rather than being divided into a mixture of
separate species and superspecies, some with disjunct distributions. From
Mexico to northern South America, I recommend recognition of the fol-
lowing allospecies: auriceps, forficatus, canivetii, assimilis, melanorhyn-
chus, gibsoni, olivaresi, and mellisugus. Together, the members of the
mellisugus superspecies show a fascinating pattern of allopatric distribu-
tions and mosaic geographical variation that bespeaks a complex evolu-
tionary history fully in keeping with the complicated geological history
of Central America and northern South America during the latter part of
the Cenozoic. These hummingbirds would seem to comprise a most fruit-
ful group for further genetic, phylogenetic, and biogeographical studies.

ACKNOWLEDGMENTS

For logistic and financial support during the expedition to the Sierra de Chiribiquete, I
am grateful to the Institute de Cooperacion Tecnica of Spain, the Institute de Ciencias
Naturales, and INDERENA. For help and companionship in the field, I thank Jose Luis
Tellen'a, Mario Dfaz, Bernardo Ortiz, Augusto Repizzo, and all the other members of our
expedition; the surgical expertise of Carlos Castano was particularly appreciated. Diego
Silva and Tomas Walschburger of the Fundacion Puerto Rastrojo kindly shared with me
their observations from the Rio Mesay area. The following kindly provided access to spec-
imens of Chlorostilbon in their care: Hno. Roque Casallas of the Museo de La Salle, Bogota;
Hernando Chirivi and Jorge Morales of INDERENA-UNIFEM; and Hno. Marco A. Serna
of the Museo del Colegio de San Jose, Medellin. I am grateful to M. Man'n for providing
measurements of specimens in the Louisiana State Museum of Natural Science. Arturo
Rodriguez contributed valuable curatorial assistance in the Institute de Ciencias Naturales.
S. N. G. Howell and M. Robbins provided useful comments on the manuscript.

LITERATURE CITED

American Ornithologists’ Union. 1983. Check-list of North American birds, 6th edition.

A. O. U., Washington, D.C.

THE WILSON BULLETIN • Vol. JOS, No. 1, March 1996

Capparella, a. 1988. Genetic variation in Neotropical birds: implications for the speciation
process. Acta XIX Intnl. Ornithol. Congr.: 1658-1664.

Chapman, E M. 1917. The distribution of bird-life in Colombia: a contribution to a bio-
logical survey of South America. Bull. Amer. Mus. Nat. Hist. 36:1—728.

Cracraft, J. 1983. Species concepts and speciation analysis. Current Ornithol. 1:159-187.

DIaz, M., E G. Stiles, and J. L. Telleri'a. 1996. Las comunidades de aves de las partes
superiores de la Sierra de Chiribiquete, Depto. del Caqueta, Colombia. Ardeola (in
press).

Estrada, J. and J. Euertes. 1993. Estudios botanicos en la Guyana colombiana V. Notas
sobre la vegetacion y la flora de la Sierra de Chiribiquete. Revta. Acad. Colomb. Cienc.
18:483-497.

Gill, E B. 1973. Intra-island variation in the Mascarene White-eye, Zosterops borbonica.
Ornithol. Monogr. 12:1—58.

Grant, P. R. 1965. The adaptive significance of some size trends in island birds. Evolution
19:355-367.

Haffer, j. 1974. Avian speciation in tropical South America. Publ. Nuttall Ornithol. Club
14:1-390.

. 1985. Avian zoogeography of the Neotropical lowlands. Ornithol. Monogr. 36:

113-146.

Hilty, S. L. and W. L. Brown. 1986. A guide to the birds of Colombia. Princeton Univ.
Press, Princeton, New Jersey.

Howell, S. N. G. 1993. Taxonomy and distribution of the hummingbird genus Chlorostil-
bon in Mexico and northern Central America. Euphonia 2:25-37.

Mayr, E. and W. H. Phelps, Jr. 1967. The origins of the bird fauna of the South Venezuela
highlands. Bull. Amer. Mus. Nat. Hist. 136:269-328.

McKitrick, M. C. and R. M. Zink. 1988. Species concepts in ornithology. Condor 90:1-14.

Meyer de Schauensee, R. 1948-1952. The birds of the Republic of Colombia. Caldasia
5(22-26):25 1-1212.

. 1960. The birds of Colombia and adjacent areas of South America. Livingston
Publ. Co., Narberth, Pennsylvania.

. 1966. The species of birds of South America with their distributions. Livingston
Publ. Co., Narberth, Pennsylvania.

AND W. H. Phelps, Jr. 1978. A guide to the birds of Venezuela. Princeton Univ.
Press, Princeton, New Jersey.

Ridgway, R. 1911. The birds of North and Middle America, part V. Bull. U.S. Natnl. Mus.,
vol. 50.

Sibley, C. G. and B. L. Monroe, Jr. 1990. Distribution and taxonomy of birds of the
world. Yale Univ. Press, New Haven, Connecticut.

Smithe, E 1975, 1981. Naturalists’ color guide. Amer. Mus. Natl. Hist. Press, New York,
New York.

Stiles, E G. 1983. Systematics of the southern forms of Selasphorus (Trochilidae). Auk
100:31 1-325.

. 1985a. Geographic variation in the Fiery-throated Hummingbird, Panterpe insig-
nis. Pp. 23-30 in Neotropical ornithology (P. A. Buckley, M. S. Foster, E. S. Morton,
R. R. Ridgely, and E G. Buckley, eds.). Ornithol. Monogr. 36.

. 1985b. Seasonal patterns and coevolution in the hummingbird-flower community
of a Costa Rican subtropical forest. Pp. 757—787 in Neotropical ornithology (P. A.
Buckley, M. S. Foster, E. S. Morton, R. R. Ridgely, and F. G. Buckley, eds.). Ornithol.
Monogr. 36.

Stiles • A NEW COLOMBIAN EMERALD HUMMINGBIRD

. 1995. Distribucion y variacion en el Ermitafio Carinegro {Phaethornis anthophilus)

en Colombia. Caldasia 18:119—130.

AND A. E Skutch. 1989. A guide to the birds of Costa Rica. Cornell Univ. Press,

Ithaca, New York.

, J. L. TellerIa, and M. Diaz. 1995. Observaciones sobre la ecologia, composicion

taxonomica, y zoogeografia de la avifauna de la Sierra de Chiribiquete, Depto. del
Caqueta, Colombia. Caldasia 17:481-500.

Wetmore, a. 1968. The birds of the Republic of Panama, part 2. Smithsonian Misc. Coll,
vol. 150, pt. 2.

Zar, J. H. 1988. Biostatistical analysis. Prentice-Hall, Inc., Englewood Cliffs, New Jersey.

Zimmer, J. T. 1950. Studies of Peruvian birds #58. The genera Chlorostilbon, Thalurania,
Hylocharis, and Chrysuronia. Amer. Mus. Nat. Hist. Novitates, no. 1474:1—12. •

, AND W. H. Phelps, Jr. 1952. Two new birds from Venezuela. Amer. Mus. Nat.

Hist. Novitates, no. 1544:1-5.

COLOR PLATE

Publication of the frontispiece painting has been made possible by an endowment estab-
lished by George Miksch Sutton.

Wilson Bull, 108(1), 1996, pp. 28-35

REPRODUCTION AND MOVEMENTS OF MOUNTAIN
PLOVERS BREEDING IN COLORADO

Fritz L. Knopf and Jeffery R. Rupert

Abstract.— North American populations of Mountain Plovers {Charadrius montanus)
have declined 63% since 1966. Using radiotelemetry, we monitored plover nesting and
brood rearing during the 1993 and 1994 breeding seasons in Weld County, Colorado. Our
objectives were to extend preliminary breeding studies of 1992 (Miller and Knopf 1993)
and to determine the minimum area required for successful reproduction by this species.
Plovers began arriving on the breeding grounds in mid-March. Due to a high incidence of
predation, eggs hatched in only nine of 34 (26%) nests in 1993 and in 20 of 54 (37%) nests
in 1994. Daily survival rates of chicks were 0.957 (442 telemetry days, 44 chicks) and 0.951
(610 telemetry days, 42 chicks) each year, respectively. Plover broods moved an average
337 ± 46.5 m/day (N = 30) and 298 ±41.9 m/day (N = 14) in 1993 and 1994 (r = 1.10,
P = 0.27), respectively. Plovers that raised chicks to fledging used between 28 and 91 ha,
averaging 56.6 ±21.5 ha. Daily movement rates (r = 0.7, P = 0.48) and total area used (/
= 1.4, P = 0.17) were similar between broods where ^ one chick fledged and broods from
which no chicks fledged. Success of plovers in raising chicks appeared related either to
overall fox activity in the area or to how effectively the adult detected and distracted foxes.
Most plovers left the breeding grounds in mid-to-late July, four months after arrival. Pop-
ulation declines of Mountain Plovers appear independent of any recent landscape fragmen-
tation within this breeding stronghold of the species. Received 10 May 1995, accepted 1
Oct. 1995.

North American Mountain Plover {Charadrius montanus) populations
declined 63% between 1966 and 1991 (Knopf 1994) and the decline has
continued through 1993 (Knopf, in press). This species nests across the
western Great Plains and eastern Colorado Plateau region, with the breed-
ing stronghold being in Weld County, Colorado (Graul and Webster
1976). Mountain Plover nests are located in microsites of native short-
grass prairie dominated by blue grama (Bouteloua gracilis) with 30% or
more of the area being bare ground (Knopf and Miller 1994; Knopf and
Rupert 1996). A female plover lays a clutch for the male to incubate then
a second clutch for herself (Graul 1973). The precocial chicks receive
uniparental care and move up to 800 m from the nest shortly after hatch-
ing (Graul 1975). Brood-rearing habitat is described as areas with forbs
or objects such as fence posts where chicks can find shade to avoid mid-
day heat (Graul 1975).

In this study, we used radiotelemetry to monitor movements and de-
termine the area requirements of Mountain Plovers during the breeding
season. We also provide information on nest success and chick survival
to supplement preliminary findings of Miller and Knopf (1993) in 1992.

U.S. National Biological Service, 4512 McMurry Ave., Fort Collins, Colorado 80525-3400.

Knopf and Rupert • MOUNTAIN PLOVERS

We interpret our findings relative to conservation of this rapidly declining
species.

METHODS

We studied Mountain Plovers during the 1993 and 1994 breeding seasons on the Pawnee
National Grassland (PNG), a 781-km^ shortgrass prairie. Graul (1973) summarized the phys-
iography, vegetation, and climate of this region. We concentrated our study on three grazing
allotments (Keota, Keota Steer, and Owens) in the vicinity of Keota, Colorado, and on three
other allotments (Reno, Sand, and Wildhorse) northwest of Briggsdale, Colorado.

Once a nest was located, we monitored the progression of incubation (29 days) by egg
flotation every 3-5 days. During nest monitoring, we drove a vehicle near the nest and,
remaining in the vehicle, reached down to remove an egg to avoid spreading human scent
in the vicinity. Within 2-A days before hatching we captured adult birds on nests with a
fishing line snare or swing-door box trap made of wire mesh. In 1993, some juveniles from
unknown nests were captured by hand and aged per Miller and Knopf (1993) to supplement
survival rate data on older chicks.

All birds were banded with a U.S. Fish and Wildlife Service numbered metal band on
one leg and colored plastic bands on the other. Plovers (N = 43) were fitted with 1.5— 3.0
g radio transmitters with 15 cm antennae (Holohil Ltd., Woodlawn, Ontario, Canada, and
Advanced Telemetry Systems, Isanti, Minnesota. Mention of commercial products does not
constitute endorsement by the U.S. Government). The transmitter was affixed by applying
a light coating of waterproof epoxy adhesive (Titan Corp., Lynnwood, Washington) and
sliding it under the upper back feathers. Care was taken not to expose the skin to the epoxy.
The transmitter was not visible on the birds and the whip antenna was difficult to see even
with binoculars at close (<20 m) distances. Transmitter life was about 90 days, and trans-
mitters remained attached to adult birds until the prebasic molt. Birds were not handled after
initial transmitter attachment.

All chicks in monitored nests were color banded on the day of hatching, then never
rehandled. We subsequently located broods almost daily at distances up to 1000 m with a
TRX-1000 Wildlife Materials Inc. (Carbondale, Illinois) receiver and a hand-held, three-
element Yagi antenna. All relocations were from a vehicle, again to avoid spreading human
scent. Specific locations of plovers were recorded using the Magellan NAV 5000 Global
Positioning System [Magellan Systems Corporation, San Dimas, California]. Position read-
ings were recorded to the nearest hundredth of a minute and only if a satellite geometric
quotient registered > 7 on a scale of 1-9. Based upon readings at a known (surveyed)
benchmark, accuracy of our specific instrument was calculated at 7.2 ± 1 .4 (x ± SD) m
latitude and 8.4 ± 1 .6 m longitude for an area error of ± 60.6 m^.

The area used by broods was estimated by superimposing a grid over a map of the study
area and counting the number of new cells (10 X 10 m) visited over time. This is an
adaptation of the grid-cell home range estimation method first proposed by Siniff and Tester
(1965). Whereas di.scontinuous telemetry data often result in successive locations being
several cells apart, we conservatively assumed that birds traveled in a straight line between
locations. Thus, area-of-use descriptions are considered to be the minimum (vs actual) area
requirements for raising a brood.

RESULTS

Egg/chick survival. — Plovers were first observed on the PNG on 17
March 1993 and 21 March 1994, although the exact date of the first bird

THE WILSON BULLETIN • Vol. 108, No. I, March 1996

Table 1

Productivity and Chick Survival Rates of Mountain Plovers on the Pawnee
National Grassland, Weld County, Colorado, 1992-1994

X No./Successful nest

Year

No.

nests

Nest

success

(%)

Eggs

hatched

Chicks

fledged

Chicks

migrated®

Daily

survival

rate

1992'=

2.6

1.21

0.74

0.977

1993

2.4

0.26

0.22

0.957*=

1994

2.6

0.35

0.17

0.95 U

“Calculated from average of 18 days that chicks remained on breeding grounds after fledging (Miller and Knopf 1993).
•’ Data from Miller and Knopf (1993).

' Ba.sed on 442 telemetry days, 44 chicks
Based on 610 telemetry days, 42 chicks.

arriving may have been a day or two earlier. We monitored 34 nests in
1993 and 54 nests in 1994. Three clutches were abandoned each year. In
1993, two clutches were abandoned after being partially predated. The
other clutch was abandoned after it had been flooded from 4 to 6 June
and then incubated again from 7 to 20 June. In 1994, two clutches were
abandoned for unknown reasons. One other was abandoned after being
incubated two weeks beyond normal hatching time. Adult plovers from
three of the six nesting efforts stayed in the nest area until migrating,
whereas the other three left the area within 48 hours and could not be
relocated.

Predation rates on both eggs and chicks were high (see Miller and
Knopf [1993] for a list of potential predators), and egg and chick survival
were low (Table 1). Survival probabilities of the 1994 chicks partitioned
into 10-day intervals (day 1-10 = 0.935, 1 1-20 = 0.957, 21-30 = 0.970,
and 31^0 = 0.971) indicate that survival increased with age of the chick.
Generally, daily survival rates were only slightly lower than reported for
1992 (Miller and Knopf 1993). However, the lower daily rates resulted
in a drastic decline in the number of chicks produced per nesting effort
if projected to the time when fledged chicks left the breeding area.

Movement patterns. — We captured and placed transmitters on 17 and
26 adult plovers with broods in 1993 and 1994, respectively. Each brood
was tracked until either all members of the brood were killed by a pred-
ator or fledged chicks left the nest vicinity. Of adults fitted with trans-
mitters, we obtained prolonged movement and area-of-use data on seven
and 14 broods in 1993 and 1994, respectively. An additional 26 chicks
were aged and fitted with transmitters in 1993, of which 23 provided
usable information. Thus, data were obtained on 30 broods in 1993 and

Knopf and Rupert • MOUNTAIN PLOVERS

14 broods in 1994. The average duration of radio-tracking was 10.3 ±
5.2 and 22.9 ± 13.4 days for each brood in 1993 and 1994, respectively.

Based upon 290 telemetry days, plover broods moved an average of
337 ± 46.5 m/day in 1993 (range 61 to 600 m). The 320 telemetry days
in 1994 indicated an average move of 298 ± 41.9 m/day (range 85 to
651 m). Distances moved were similar (r = 1.10, P = 0.27, 608 df)
between years. Of the 42 broods, 38 moved 100—500 m/day. Of the re-
maining four broods, two were monitored for only five days and moved
<100 m/day. Two others, monitored four days and 13 days, moved 1000
and 1085 m/day, respectively.

Most losses of chicks were to swift foxes (Vulpes velox). Average daily
movements were similar {t = 0.7, P = 0.48, df = 40) among broods in
which all chicks were killed by predators (400 ± 313 m, N = 15) and
broods where chicks fledged (340 ± 226 m, N = 27). The total distances
moved during the first 10 days after hatching of eggs (when predation
rates were highest) were also similar {t - 0.50, P - 0.62, df = 14)
between broods in which all chicks were lost and broods where chicks
fledged.

Brood-rearing area. — Due to the high rates of predation and some
transmitters on older chicks becoming dislodged, we were only able to
track six broods from hatching to fledging. The total minimum area used
by those six broods ranged from 28 ha to 91 ha and averaged 56.6 ±
21.5 ha. As suspected from the movement data, the average area used on
a daily basis by broods where chicks fledged (2.5 ± 1.6 ha, N = 27) vs
broods that lost all chicks to predators (3.4 ± 2.4 ha, N = 15) were
similar {t = 1.4, P = 0.17, df = 40).

Timing of departure. — At least two adults renested after losing a clutch
or brood early in the breeding season (May, early June). One adult whose
chicks hatched on 21 May 1993 was located on a second nest 140 m
away on 21 June. The transmitter had failed, so knowledge of the fate of
the first brood of three chicks is not certain. This bird abandoned its
second clutch after it was partially depredated; it remained in the vicinity
with other plovers until 14 July when it left the area, returned 26-29 July,
then left the area for the season.

This example illustrates the general pattern of flocking and departure
from the breeding grounds. A few adult plovers began congregating in
flocks in mid-June each year. Fledglings started to appear in these flocks
in July. Many adults also were undergoing a prebasic molt at this time,
precipitating an increased rate of transmitter loss. Plovers with transmit-
ters began leaving the study areas gradually after mid-June, but a major
exodus of plovers occurred mid-late July. Seventeen plovers in 1993 and
eight plovers in 1994 which were being located daily were known to have

THE WILSON BULLETIN • Vol. 108, No. I, March 1996

left the study area between 14 July and 3 August. In 1993, an aerial
survey of Weld County from 3 to 9 August confirmed that not one of 17
transmittered birds remained in the vicinity. The few adults (transmittered
and untransmittered) that we could find in early August of 1993 all had
large chicks incapable of flight.

DISCUSSION

Productivity. — Miller and Knopf (1993) reported that adult survival
and productivity on the PNG were similar to historical studies. We re-
cently documented that survival rates of birds wintering in California are
also high (Knopf and Rupert 1995). In this study, we recorded daily
survival rates of chicks in 1993 and 1994 similar to those reported for
1992. The more intensive efforts described here detected an increasing
probability of survival with age of the chick, as reported by Graul (1975).
Thus, the 1992 daily survival rate (0.977) was inflated slightly by a high
incidence of older chicks being transmittered that year.

An unknown percentage of female plovers lay two clutches, the male
incubates the first as the female lays a second clutch for herself (Graul
1973). The incidence of male-incubated clutches may increase with food
abundance (Graul 1976). Thus, whereas 0.26 and 0.35 chicks fledged per
nest compares poorly with fledging rates of some other North American
plovers (Page et al. 1983, Haig and Oring 1988, Prindiville Gaines and
Ryan 1988), the actual productivity per pair may be up to twice that value
in a given year.

The number of eggs hatched per clutch has remained stable when com-
pared to studies conducted 10 (McCaffery et al. 1984) and 25 (Graul
1975) years earlier. Most reproductive losses were due to fox predation.
We believe that the greater losses of nests and chicks to predators in 1993
and 1994 compared to 1992 were attributable indirectly to reduced food
resources. Mountain Plover reproductive efforts (Graul 1976), as those of
other grassland birds (George et al. 1992), are less successful in years of
drought. The two seasons of this study were drought years on the PNG.
Grasshoppers are a major food item of both plovers and foxes, and their
populations were very low both years. Low grasshopper populations
would increase the time spent foraging for both species, thus increasing
the probability of eggs and chicks being detected by foxes.

Movement patterns. — Reproductive success of many birds appears to
be a tradeoff between acquiring food and risking predation (Martin 1992).
From an evolutionary perspective, numerous traits of Mountain Plovers
may reduce detection by predators, including the cryptic coloration of
chicks and crypsis to avoid detection (Sordahl 1991), two clutches in-
cubated separately by the two adults (Graul 1973), shell removal at hatch-

Knopf and Rupert • MOUNTAIN PLOVERS

ing (Graul 1975), rapid movement of chicks away from the nest (Graul
1975), predator distraction displays by adults (McCaffery et al. 1984),
and the ability of chicks to fly at only 70% of adult body weight (Miller
and Knopf 1993).

Mountain Plovers led hatchlings away from the nest as soon as they
were dry. We regularly tracked the directional movement of a brood up
to 2 km within two or three days of hatching. Many plovers moved broods
to areas of disturbed prairie (Knopf and Rupert 1996) and then remained
in those general areas. In contemporary prairie landscapes, such distur-
bances are either areas frequented for watering and loafing by cattle or
fallow agricultural fields. After the initial move, many broods remained
in the vicinity of these areas where chicks foraged on small insects.

Area requirements. — This study was precipitated by the conservation
need for information on the minimum area necessary for plovers to raise
chicks. The minimum area within which a brood was raised was 28 ha.
Plovers raised chicks in broadly overlapping areas, with two or three
broods sometimes occurring in a general vicinity (such as around a cattle
watering tank). Thus, the potential exists for a suitable area to meet the
needs for more than one bird to raise chicks successfully.

The movement rates and area-of-use by plovers that successfully raised
their chicks to fledge did not differ from those of plovers that lost the
entire brood. Swift foxes were often seen hunting during daylight hours
(especially after mid-June as pups began eating prey) in addition to night
foraging. The eventual success of a plover in raising its chicks appeared
to be the result of either the overall fox activity in the immediate area
being used by the brood or the effectiveness of the adult plover in de-
tecting and distracting a fox.

Departure from breeding grounds. — Most Mountain Plovers that bred
in Weld County left by 1 August each year. Occasional flocks of plovers
were seen after early August, with the latest being on 8 October 1993.
We assumed the later flocks were of birds moving south from more north-
erly nesting areas. The first plovers arrive on California wintering grounds
in mid-October. The prolonged period of migration to the wintering areas
is in sharp contrast to the apparently direct, nonstop flight from California
back to Colorado in March (Knopf and Rupert 1995). Plovers spend ap-
proximately four months on their Colorado breeding grounds, five months
on the California wintering grounds, and three months moving from the
former to the latter.

CONCLUSIONS

Miller and Knopf (1993) concluded that recent population declines of
Mountain Plovers can be attributed either to long-term declines in repro-

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

ductive success or to phenomena occurring at nonbreeding areas. The
comparatively lower survival rates of chicks observed in 1993 and 1994
support that preliminary conclusion. Although swift fox densities seem
high on the PNG, the fox is only locally distributed within the breeding
range of the plover. We would not expect, for example, such a high rate
of predation in Phillips County, Montana (the second major breeding
locale) due to the absence of foxes in that state.

The shortgrass prairie, like many native vegetative associations, has
become highly fragmented in the last century. The PNG is no exception,
with 781 km^ occurring in 130 parcels ranging from 16 to 23,895 ha (B.
Ladd, pers. commun.). Although some of the smaller PNG parcels are
<28 ha, the minimum area required for brood rearing, all are contiguous
to private lands that are likewise managed as rangelands. We feel that
these private lands also provide suitable Mountain Plover habitat. Thus,
we conclude that virtually all relatively flat, grazed shortgrass prairie par-
cels provide potential habitats for plovers on the PNG. The current decline
of the North American Mountain Plover population appears independent
of recent fragmentation of landscapes within the native shortgrass prairie
of northeastern Colorado, the breeding stronghold (Graul and Webster
1976) of the species.

ACKNOWLEDGMENTS

We thank the U.S. Lorest Service, for financial assistance, and personnel of the Rocky
Mountain Region, Arapaho-Roosevelt National Lorest, and the Pawnee National Grassland
District for technical assistance. We are especially grateful to Larry Mullen for administra-
tive support. The Colorado Division of Wildlife conducted aerial reconnaissance, and we
specifically thank Jim Dennis for his efforts. Susan Skagen and Tex A. Sordahl provided
helpful comments on the manuscript. Clif Knopf provided assistance in the field.

LITERATURE CITED

George, T. L., A. C. Fowler, R. L. Knight, and L. C. McEwen. 1992. Impacts of a severe
drought on grassland birds in western North Dakota. Ecological Applications 2:275—
284.

Graul, W. D. 1973. Adaptive aspects of the Mountain Plover social system. Living Bird
12:69-94.

. 1975. Breeding biology of the Mountain Plover. Wilson Bull. 87:6—31.

. 1976. Food fluctuations and multiple clutches in the Mountain Plover. Auk 93:

166-167.

AND L. E. Webster. 1976. Breeding status of the Mountain Plover. Condor 78:

265-267.

Haig, S. and L. W. Oring. 1988. Mate, site, and territory fidelity in Piping Plovers. Auk
105:268-277.

Knopf, F. L. 1994. Avian assemblages on altered grasslands. Stud. Avian. Biol. 15:247—
257.

. 1996. Prairie legacies — birds in Prairie conservation: preserving North America’s

Knopf and Rupert • MOUNTAIN PLOVERS

most endangered ecosystem (E B. Samson and E L. Knopf, eds.)- Island Press, Covelo,
California (in press).

AND J. R. Rupert. 1995. Habits and habitats of Mountain Plovers in California.

Condor 97:743-751.

AND . 1996. Declining species on private lands: Mountain Plovers on plowed

ground. Wildl. Soc. Bull, (in press).

Martin, T. E. 1992. Interaction of nest predation and food limitation in reproductive suc-
cess. Current Ornithol. 9:163-197.

McCaffery, B. J., T. a. Sordahl, and P. Zahler. 1984. Behavioral ecology of the Moun-
tain Plover in northeastern Colorado. Wader Study Group Bull. 40:18—21.

Miller, B. J. and E L. Knopf. 1993. Growth and survival of Mountain Plovers. J. Field
Ornithol. 64:500—506; 65:193.

Page, G. W., L. E. Stenzel, D. W. Winkler, and C. W. Swarth. 1983. Spacing out at
Mono Lake: breeding success, nest density, and predation in the Snowy Plover. Auk
100:13-24.

Prindiville Gaines, E. and M. R. Ryan. 1988. Piping Plover habitat use and reproductive
success North Dakota. J. Wildl. Manage. 52:266—273.

SiNiFF, D. B. and j. R. Tester. 1965. Computer analysis of animal movement data obtained
by telemetry. BioScience 15:104—108.

Sordahl, T. A. 1991. Antipredator behavior of Mountain Plover chicks. Prairie Nat. 23:
109-115.

Wilson Bull., 108(1), 1996, pp. 36-52

TRIGEMINAL REPELLENTS DO NOT PROMOTE
CONDITIONED ODOR AVOIDANCE IN
EUROPEAN STARLINGS

Larry Clark

Abstract. — Birds, and in particular European Starlings (Sturnus vulgaris), avoid con-
sumption of fluid and food treated with the natural plant products, methyl anthranilate and
o-aminoacetophenone. Avoidance is an unlearned response most likely mediated via chem-
ically sensitive fibers of the trigeminal nerve. The trigeminal nerve codes for chemical
irritation and pain. Starlings are not repelled by the odor of the compounds, nor is olfaction
important in the avoidance response. Moreover, starlings fail to learn to avoid the odor of
the repellents, even after direct oral contact with liquid repellent. While trigeminal irritants
can be powerful repellents, the absence of associative learning for these repellents will
influence the application strategy for formulation and use. More broadly, the difference in
learning abilities associated with trigeminal repellents and those commonly responsible for
conditioned avoidance learning have implications for the structure of chemical defenses of
fruits and the prevention of untimely frugivory. Received 24 Feb. 1995, accepted I Sept.
1995.

Nonlethal bird repellents are important components of an integrated
wildlife management strategy. Repellents can be used to protect birds
from human activities (Clark and Shah 1993) or to minimize damage
caused by birds (Mason and Clark 1992). The social emphasis on safe,
nonlethal methods to resolve conflicts between humans and birds has
resulted in numerous attempts to identify new repellents (Dolbeer 1986,
Crocker and Perry 1990, Clark and Shah 1991). However, reported effi-
cacy of nonlethal repellents is highly variable (Mason and Clark 1992).
In part, this is due to a misunderstanding about how repellents work. To
minimize failure rates in the field, several fundamental questions about
mode of action and formation of avoidance response remain to be re-
solved.

Nonlethal chemical repellents operate via two distinct mechanisms (Za-
horik 1976), conditioned avoidance and nonlearned avoidance. In con-
ditioned avoidance learning, birds learn to avoid sensory cues paired with
a stimulus that causes illness (Garcia et al. 1966). The magnitude and
persistence of the avoidance response depends on the toxic potential of
the sickness producing agent and the localization of the illness. Pelchat
et al. (1983) found conditioned avoidance was strongest in the rat when

United States Dept, of Agriculture, Animal and Plant Health Inspection Service, Animal Damage Control,
Denver Wildlife Research Center and Monell Chemical Senses Center, 3500 Market Street, Philadelphia,
Pennsylvania 19104. (Present Address; United States Dept, of Agriculture, Animal Plant Health Inspection
Service, National Wildlife Research Center, 1716 Heath Parkway, Fort Collins, Colorado 80524).

Clark • BIRD REPELLENTS

sickness was localized in the upper region of the small intestine. In birds,
ingestion of carbamate insecticides (e.g., methiocarb) and fungicides (e.g.,
thiram, ziram) causes gastrointestinal sickness. Substances causing sick-
ness have been used to condition birds to avoid tastes (Schuler 1983),
odors (Clark and Mason 1987), and visual cues (Mason and Reidinger
1983). In a nonlearned avoidance response, substances possess taste,
smell, or irritating qualities that are perceived as unpalatable by birds.
Generally, in the quantities ingested, these substances do not cause sick-
ness (Clark and Mason 1993).

Previous studies indicated that acetophenone and anthranilate bird re-
pellents must be present in high concentrations to be effective (Clark et
al. 1991). High concentrations of repellents (hundreds to thousands ppm)
can be delivered orally, in food or fluid, or to the eye via aerosols. Re-
sponsiveness to only high concentrations suggests mediation by the tri-
geminal system as opposed to olfaction or taste (Walker et al. 1986).
Chemically sensitive fibers of the trigeminal nerve mediate response to
irritating and painful stimuli (Green et al. 1990). Yet coding for pain or
irritation does not necessarily imply tissue damage (Clark 1995).

Common questions are whether the odors of acetophenones and an-
thranilates are repellent or whether birds can learn to avoid the odor of
these repellents. These questions imply avoidance behavior mediated via
olfaction and the formation of a conditioned avoidance response. Deter-
mining whether avoidance is influenced by olfaction or trigeminal cues
is critical to the conceptualization and implementation of delivery strat-
egies of repellents and to the understanding of how birds may respond to
natural plant or insect chemical defenses.

The experiments in this paper address these question and are used to
argue the point that acetophenone and anthranilate compounds are trigem-
inal irritants (repellents). As a test of the appropriateness of the experi-
mental paradigm, the effect of short-term water deprivation on subsequent
drinking assays was evaluated in Experiment 1. Experiment 2 tested
whether naive starlings avoided the odor of a repellent and whether oral
contact with a repellent was a sufficiently adverse experience to train
starlings to subsequently avoid the odor of a repellent. This experiment
also tested the effect of stimulus sequence on the outcome of the drinking
assays. Experiment 3 tested the effect of prolonged exposure to orally
delivered repellents on the subsequent response to the odor of the repel-
lent. In Experiment 4, the role of olfaction in the mediation of the avoid-
ance response was assessed.

METHODS

Study subjects. — European Starlings (Sturnus vulgari.<;) were decoy-trapped at Sandusky,
Ohio, and transported to the laboratory in Philadelphia where they were kept in group

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

housing until selected for experimentation. Starlings were maintained on chick starter mash
{ad libitum) supplemented with a vitamin mixture and fresh apples (weekly). Tap water was
available continuously, except during testing. Because starlings exhibit a seasonality in their
olfactory ability (Clark and Smeraski 1990), all tests were conducted during the spring when
starlings have good olfactory acuity and discrimination ability. Starlings were maintained
on a constant temperature (23°C), 14:10 h light: dark cycle during their residence in the
laboratory.

Test stimuli. — Fluid intake for o-aminoacetophenone (OAP) and methyl anthranilate (MA)
was evaluated using standard 6-h one-bottle (no-choice) drinking tests (Clark and Shah
1994). These compounds were selected as representative of nonlethal acetophenone and
anthranilate bird repellents. Both compounds are natural plant metabolites that have organ-
oleptic characteristics that humans perceive as a musky/foxy odor for OAP and a grapey
odor for MA (Acree et al. 1990). Concentrations selected for testing were based upon water
solubility limits of the least soluble compound (MA) and the least practical concentration
yielding reliable rejection of treated fluid by most starlings.

Experiment 1 : Ejfects of water deprivation on fluid consumption. — The objectives of the
first experiment were to (1) determine the effect of diurnal water deprivation on overnight
water consumption and (2) determine the effect of a three-day diurnal water deprivation
schedule on post-test water intake. Determining the potential magnitude of the carry-over
effect attributable to water deprivation was important for the interpretation of other exper-
iments relating to the effects of chemical repellents. While previous studies showed that the
6-h drinking test used here is not sufficient to cause a severe water deficit in starlings (Clark
and Shah 1991), it was decided that a revisitation of these questions was in order.

The experimental design consisted of a standard one-bottle, 6-h assay. On the first day
(the pre-test period), 18 birds were randomly assigned to three groups (N = 6/group) and
presented with tap water in graduated Richter tubes. The recording period began at 10:00
and ended at 16:00 (hereafter defined as the diurnal test period). The Richter tubes were
replaced with a second set of tubes and water was available ad libitum through the period
16:00 to lights out at 19:00, throughout the night, and for the period from lights on at 7:00
to 10:00, the start of the next test sequence (hereafter defined as the overnight period). As
a precondition for further testing, similarity for average diurnal water consumption among
groups was verified using a one-way fixed effects analysis of variance (anova).

On the second day, birds within the groups were presented with one of three randomly
assigned treatments. Birds within the control group were presented with tap water. Birds
within the second group were presented with a 28 mM solution of o-aminoacetophenone
(OAP). Because OAP is a potent bird repellent, it was anticipated that this group would
experience voluntary water deprivation by avoiding the treated water (Clark and Shah 1991).
Birds within the third group were water deprived by physically excluding them from the
Richter tubes. The protocol for recording water consumption for the diurnal and overnight
periods followed that described above. Birds retained their water presentation treatment
assignments for days three and four, and water intake for the diurnal and overnight periods
followed the format described above.

On the fifth day (the post-test period), birds within all three treatment groups were pre-
sented with untreated tap water and intake was monitored according to the prescribed pro-
tocol.

The first question, “. . . does diurnal fluid deprivation affect overnight water consump-
tion?”, was addressed using a 3 X 3 repeated measures fixed-effect analysis of covariance
(ancova). The dependent variable was overnight water consumption. The previous day’s
diurnal fluid intake was used as a covariate. Because the covariate was measured each day
along with the dependent variable, it was treated as a changing covariate, i.e. separate

Clark • BIRD REPELLENTS

residuals were calculated for each day. Days were treated as a repeated measure, and group
was treated as a between measures effect. Multivariate criteria (Rao R, Wilk’s lambda) were
used to simultaneously test repeated measures contrasts on adjusted means when there were
more than two levels on the repeated measure. This approach is to be preferred because it
does not rely on the assumption that the repeated measures are independent. In circumstances
where only univariate statistics were estimable, i.e., one degree of freedom, the F statistic
was used to assess significance. The latter case was most common in planned comparison
of treatment levels by contrasts. All post-hoc tests were analyzed using a Scheffe’s test. This
and subsequent analyses were carried out using the STATISTICAL (1994) software pack-
ages (MANOVA procedures). This software is particularly suited to repeated measures and
other longitudinal designs.

The second question, “. . . did the intervening treatments affect post-test water intake
relative to the pretreatment period?” was addressed using a 2 X 3 repeated measures fixed-
effect ANOVA. The dependent variable was diurnal fluid intake, period (pre vs post test)
was a within measures (repeated) effect, and group (i.e., water deprivation schedule) was a
between measures effect.

Experiment 2: Effects of odor and treatment sequence on water consumption. — Starlings
readily form conditioned odor avoidance responses when the unconditional stimulus pro-
duces a strong gastrointestinal illness (Clark and Mason 1987). Starlings also avoid some
substances upon initial contact, indicating that the avoidance in not learned (Clark and
Mason 1993). This latter response indicates a different mechanism for avoidance, one that
suggests hedonic attributes (e.g., taste, smell, irritation) form the basis of acceptance or
rejection. Clark and Mason termed these compounds sensory repellents. The objectives of
this experiment were to (1) determine whether oral and/or nasal exposure to GAP influenced
fluid intake, (2) determine whether starlings could learn to avoid the odor of GAP once they
had an aversive oral exposure to the substance, and (3) assess whether the order of presen-
tation of GAP in solution or odor influenced the behavioral response.

Thirty-six starlings were randomly assigned to six groups of six birds each. Gn the first
day, birds were tested for baseline water consumption in a standard 6-h drinking test fol-
lowing procedures outlined in Experiment 1. Attached to both sides of the Richter tube
sipping port was a 40 mm diameter, opaquely screened polypropylene disk containing a
wick saturated with tap water.

During the three-day test period, birds were presented with one of three treatment con-
ditions (Richter tube/wick disk combinations): [w/w] a tap water filled Richter tube and a
tap water soaked wick, [o/w] a Richter tube filled with 28 mM GAP solution and tap water
soaked wick, and [w/o] a tap water filled Richter tube and an GAP .soaked wick. Presen-
tations of the Richter tube/wick combinations to experimental cohorts were established ac-
cording to two possible Latin square designs (Table 1). Gn the fifth day (post-treatment
period), all birds were presented with the control tube/wick pairing (w/w). Diurnal and
overnight fluid intake were monitored following procedures outlined in Experiment 1.

Data were analyzed using a fixed-effects 3X3X6 nested ANGVA design with repeated
measures. Diurnal fluid intake was the dependent variable and day was a repeated measure
with three levels. Sequence (two levels) and group (six levels) were between measures
effects, with group nested within sequence (Table 1 ). To te.st whether the context of stimulus
presentation is important for repellency, contrasts between w/w and o/w, and w/w and w/o
were made for the first day of testing. These comparisons addressed the question of whether
GAP in solution (where olfactory, gustatory, and trigeminal systems may influence respond-
ing) or whether GAP as an odor stimulus (where olfactory and trigeminal systems may
influence responding) were important in the formation of an avoidance response. By con-
sidering only the first day of testing, these compari.sons controlled for possible learning

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Table 1

Latinized Designs for the Presentation Order of Treatments®

Design 1

Design 2

Day

Group 1

w/w

o/w

w/o

Group 4

w/o

o/w

w/w

o/w

w/o

w/w

o/w

w/w

w/o

w/w

o/w

w/w

w/o

o/w

“ Experimental codes are: (w/w) tap water presented in the Richter tube with blank odor disks on either side of the
drinking port; (o/w) 28 mM OAP solution in the Richter tube with blank odor disks on either side of the drinking port;
(w/o) water presented in the Richter tube with saturated OAP wicks in the odor disks on either side of the drinking port.

effects while considering whether repellency is mediated via nasal/ocular (i.e., volatile) or
oral/nasal/ocular (i.e., contact and volatile) mechanisms. Similar contrast analyses comparing
w/w to w/o were performed for days two and three. These tests allowed for the possibility
of learning while addressing the question whether consumption for birds exposed to odor
differed from that of the control condition.

OAP is a good sensory repellent when delivered orally. However, can oral presentation
of OAP be used to condition birds to avoid the odor of OAP? This question was addressed
by inspection of contrasts of the day X treatment interaction term. Intake for birds presented
with treatment levels w/w and w/o on day 2 were compared for those birds presented with
o/w on day 1, i.e., the putative training day. The overall sequence effect throughout the test
was determined by consideration of the main effect and by inspection of post hoc differences
among means grouped by treatment category using the Scheffe’s test.

Experiment 3: The conditioned odor avoidance response. — In the absence of water de-
privation effects (Experiment 1), sequence effects (Experiment 2), and an apparent inability
for short exposure to orally delivered OAP to act as an unconditional stimulus (Experiment
2), the length of time starlings were orally exposed to a chemical repellent was increased
to determine whether a conditioned avoidance response towards odor could be attained.

Separate experiments on two bird repellents were carried out, and the experimental design
of these tests was as follows. Starlings were tested serially for fluid intake in a standard no-
choice (one-bottle) 6-h drinking assay. On the first day, eighteen starlings were drawn at
random from the group housing pool, randomly assigned to one of two groups (N = 9/
group) and pre-test water consumption was monitored according to methods described in
Experiment 1 . Attached to both sides of the Richter tube sipping port was a 40 mm diameter,
opaquely screened polypropylene disk containing a wick saturated with tap water (w/w,
following the nomenclature convention in Experiment 2).

Eor each of three successive days (2-4), starlings within groups were presented with
Richter tubes containing a 28 mM solution of a sensory repellent (OAP or MA) and diurnal
fluid intake was monitored as described above (Experiment 1). Attached to both sides of
the Richter tube sipping port was a wick saturated with tap water (o/w). The strong odor
derived from the repellent emanated from the drinking port of the Richter tube, i.e., repellent
solution.

On the fifth and sixth day, starlings were presented with Richter tubes containing tap
water and fluid intake was monitored as above. During this period the wick inside the disks
was saturated with repellent (w/o). Thus, the strong odor of the repellent was present in the
apparatus but did not originate from the .solution.

Data were analyzed using a fixed effects one-way ANOVA with repeated measures on

Clark • BIRD REPELLENTS

days/treatment effect. The confounding of day and treatment on water intake was considered
unimportant. Thus, in the absence of carryover effects due to diurnal water deprivation
(Clark and Shah 1991, Experiment 1), any day/treatment effect was to be interpreted pri-
marily as a treatment effect. In addition to the day/treatment effect, a single a prion planned
comparison was made using contrasts. The w/w condition on day 1 was compared to the
w/o condition (days 5-6) as an assessment for the formation of a conditioned odor avoidance
response. The results for the GAP and MA tests were analyzed separately.

Experiment 4: The role of olfaction in the avoidance response. — Odors can stimulate the
olfactory system, but they can also gain access to chemically sensitive receptors of the
trigeminal system via the mouth, nose, or eye. Thus, avoidance of an odor might be based
upon its unpleasant smell or on its irritating/painful properties.

The object of Experiment 4 was to determine the influence of olfaction on fluid intake
when odor was present in the fluid to be consumed. The experimental design was as follows.
Starlings were tested for fluid intake in a standard no-choice (one-bottle) six hour drinking
assay. The experiment comprised two surgical conditions. (Surgery effect: sham or bilateral
olfactory nerve cut [BONG]), two repellent concentration conditions (Concentration/Group
effect: 14 mM and 28 mM), and two test periods (day/treatment effect: pretreatment = water
presentation and treatment = chemical repellent presentation). The confounding of time and
treatment was considered to be unimportant (Experiment 1), and any differences were as-
sumed to be attributable to the presence of repellent. Data were analyzed using a fixed
effects, 3-way analysis of variance with repeated measures on the day/treatment effect.

Sixteen starlings were selected at random from the group housing pool and randomly
assigned so that eight received a BONG and the other eight received a sham surgery.
Starlings were anesthetized with choral hydrate/penabarbitol at a dose of 2 ml/kg, intra-
peritoneally and placed in a head-holder. Surgery for BONG and sham treatments followed
procedures outlined in Glark and Mason (1987). Following surgery, starlings were housed
individually in cages (61 X 36 X 41 cm). Two weeks following surgery, birds were adapted
to experimental conditions. The two-week latency period was estimated to provide sufficient
time for degeneration of olfactory afferents into the olfactory bulb, yet not sufficiently long
to allow olfactory nerve regeneration (Wenzel and Salzman 1968). Following adaptation
one half of the birds within the BONG and sham surgery treatments was randomly assigned
to concentration groups, resulting in four birds per experimental cell. On the first day of
testing (pretreatment period), beginning at 09:30 h, consumption of tap water was recorded
every 2 h for the next 6 h. On the second day, starlings were presented with their preassigned
concentration of OAP and fluid intake was monitored every 2 h for the next 6 h. Fluid was
presented in 120 ml graduated Richter tubes. Starlings were visually isolated from one
another as well as from the contents of the Richter tubes. After five days rest, the same
birds used in the above experiment were re-randomized with respect to concentration group
assignment and tested with MA using the above protocol.

At the end of the experiments, birds were killed with pentobarbital and sequentially
perfused with physiological saline, 10% formalin and a 10% formalin and 30% sucrose
mixture. Brains were removed and the effect of the nerve section on olfactory bulb structure
was evaluated in a double blind sequence. Briefly, the anterior tips of the hemispheres
(containing the olfactory bulbs) were cut into 50 p-m slices. Every fifth slice was stained
with Nissl stain to highlight the outline of the glomeruli. The observer (familiar with normal
histological structure of avian olfactory bulbs) categorized the coded histological series as
being either in a degenerative state or normal. Goncordance of the observer s scoring and
the surgical status were compared using a chi-square analysis. Degeneration of the glomer-
ular structure was taken as evidence for lack of olfactory nerve input into the olfactory bulb
(Wenzel and Salzman 1968).

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

1 2 3

COE COE COE

10 -

5 -

Night

Eig. 1 . Average overnight water consumption as a function of time and water deprivation
schedule (Experiment 1). Open circles = control group (C); solid circles = OAP group (O);
Shaded circles = water excluded group (E). Horizontal bars depict statistically similar water
consumption (Scheffe s test, P > 0.05). Letters compare average intake across treatment
categories within a single night. Numerals compare average intake across nights within a
single treatment category. Vertical capped bars depict ± one standard error.

Zinc sulfate (ZnS04) also has been used to render birds anosmic. However, because ZnSOj
strips away the olfactory epithelium (Burd 1993), it may also affect epithelial layers con-
taining trigeminal free nerve endings. There are no data on this latter point. Because of this
uncertainty, surgical manipulation of the olfactory nerve was deemed to be the best way to
determine the role of olfaction in the avoidance response.

Directly eliminating trigeminal input is not feasible. In birds, only the ophthalmic branch
of the trigeminal nerve (OBTN) is easily accessible for denervation studies (Getty 1975).
Manipulation of the maxillary and mandibular branches would require too much destruction
of bone and muscle tissue. All three branches could be eliminated at the Gasserian ganglion
located in the eye orbit, but this would require permanent blinding of the bird.

RESULTS

Experiment I : Effects of water deprivation on fluid consumption. — The
overnight fluid intake for the treatment groups, adjusted for the covariate
of diurnal fluid intake, differed across days (Rao r = 4.55, P < 0.007).
The post-hoc analysis indicated that the overnight water consumption
within treatment groups was similar across nights (Fig. 1). However, water
consumption differed among treatments within nights. On the first night,
all birds consumed similar amounts of water. On the second and third

Clark • BIRD REPELLENTS

Fig. 2. Average diurnal water consumption as a function of treatment group and time
(Experiment 1). Open circles = control group; solid circles = OAP group; shaded circles
= water excluded group). Vertical capped bars depict ± one standard error.

nights the control birds drank less water than the birds that were physi-
cally excluded from water during the day. Overnight consumption for
birds presented with OAP did not differ from the other two experimental
groups. There was an inverse, albeit nonsignificant (P < 0.099), relation-
ship between daytime fluid intake and subsequent overnight water con-
sumption (Fig. 1). Birds experiencing fluid deprivation by exclusion tend-
ed to have the highest overnight water intake (36.0 ml ± 3.7 SE), fol-
lowed by birds experiencing deprivation by repellency (30.4 ml ± 3.6
SE). Control birds that had ad libitum access to water showed the lowest
overnight water consumption (24.3 ml ± 2.7 SE). Treatment category did
not appear to affect baseline diurnal water consumption immediately fol-
lowing the tests (Fig. 2). There were no group {P < 0.453), test period
{P < 0.889), or interaction (P < 0.362) effects.

Experiment 2. — The context of the stimulus is important for repellency.
With no prior experience, starlings presented with OAP in solution con-
sumed less fluid than the controls (P = 94.86, df = 1,30, P < 0.001).
This observation stands in contrast to the similarity of diurnal fluid intake
for the controls (w/w) and birds presented with water and OAP odor (w/
o) (P < 0.45). Thus, odor alone is not a sufficiently strong stimulus to

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Test Day 113232 131223 121323

Test Group 614325 265143 463521

Eig. 3. Average, diurnal fluid intake as a function of time and treatment (Experiment
2). Birds in the control group (w/w, open bars) were presented with tap water in the drinking
tubes and disks contained water soaked wicks, birds in the OAP group (o/w, solid bars)
were presented with a 28 mM solution of o-aminoacetophenone in the drinking tubes and
disks contained water soaked wicks, birds in the odor group (w/o, hatched bars) were pre-
sented with water filled drinking tubes and wick soaked in OAP. Average intake is ranked
within treatment type and nonsignificant {P > 0.05) differences among means are depicted
by the horizontal bar. Test Day and Test Group are the column and row label designations
of the two Latinized presentation designs described in Table 1 . Vertical capped bars depict
one standard error.

result in avoidance. This pattern persists even for experienced birds. On
days two and three the comparison w/w to w/o indicates similar levels of
fluid intake {P < 0.966 and 0.21, respectively), irrespective of the pre-
vious treatment exposure. The lack of an overall test sequence {P <
0.305) or day {P < 0.121) effect on diurnal fluid intake is apparent in
Fig. 3.

Starlings do not readily acquire a conditioned avoidance response to
the odor of OAP when orally delivered OAP is used as the unconditional
stimulus. Comparison by contrasts of w/w to w/o on the second day of
testing when both groups received o/w on the first day failed to uncover
differences in fluid consumption {P < 0.591).

As was the case for Experiment 1, the intervening three day treatment
schedule had no effect on baseline water consumption. There were no
group, day or interaction effects for the pre-test vs post-test comparison

Clark • BIRD REPELLENTS

■4—'

■TD

Treatment Day

Fig. 4. Average fluid intake for OAP as a function of time and solution treatment (Ex-
periment 3). The left most open bar depicts intake of tap water paired with a water soaked
wick in the odor disc (w/w). The middle solid bars depict fluid intake of a 28 mM o-
aminoacetophenone (OAP) solution paired with a water soaked wick in the odor disc (o/
w). The right most hatched bars depict fluid intake of tap water paired with an OAP soaked
wick in the odor disc (w/o). Vertical capped bars depict one standard error.

of diurnal water consumption (P < 0.351, < 0.434, < 0.674, respec-
tively).

Experiment 3: The conditioned odor avoidance response. — There were
significant day/treatment effects for both the OAP and MA tests (Rao’s r
= 9.56, P < 0.024 and r = 12.48, P < 0.15, respectively). The planned
comparisons indicated that fluid consumption was similar for the w/w and
w/o treatment conditions for both OAP (P < 0.72, Fig. 4) and MA (P <
0.87, Fig. 5). Thus, three days exposure to orally administered repellent
was not sufficient for the formation of a conditioned avoidance response
to that odor.

Experiment 4: The role of olfaction. There was a strong day/treatment
effect for OAP (P = 74.57, df = 1,12, P < 0.001), showing that overall
fluid intake was suppressed when the starlings were presented with either
28 or 14 mM OAP solutions (Fig. 6). None of the other effects achieved
probability levels less than P = 0.15. The planned comparisons between
sham-OAP and BONC-OAP within each concentration showed that sur-
gery had no effect on avoidance of OAP (P < 0.619, 0.62, respectively).

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

^ 25

^ 20
ro

^ 10
5
0

0 1 2 3 4 5 6 7

Treatment Day

Fig. 5. Average fluid intake for MA as a function of time and solution treatment (Ex-
periment 3). The left most open bar depicts intake of tap water paired with a water soaked
wick in the odor disc (w/w). The middle solid bars depict fluid intake of a 28 mM methyl
anthranilate (MA) solution paired with a water soaked wick in the odor disc (o/w). The right
most hatched bars depicted fluid intake of tap water paired with a MA soaked wick in the
odor disc (w/o). Vertical capped bars depict one standard error.

Similar consumption patterns were observed for the MA trials. There
was a strong period effect for MA (F 83.14, df = 1,12, P < 0.001),
showing that fluid intake was suppressed across all surgery and concen-
tration levels when starlings were presented with MA solutions (Fig. 7).
No other effect achieved probability levels less than P = 0.22. The
planned comparisons between sham-MA and BONC-MA within each
concentration showed that surgery had no effect on avoidance of MA (P
< 0.215, 0.399, respectively).

Post-mortem visual inspection of the glomerular layer of sham surgery
birds showed a diffuse pattern seen in normal starlings, typical for the
spring breeding season. In contrast, BONC birds showed a complete break
down of the glomerular structure. The birds were euthanized 20—25 days
after surgery. The observer scoring the slides, but blind to the experi-
mental identity, categorized all BONC tissue as having degenerative glo-
meruli (N = 8) and all SHAM tissue as being normal (N = 8) =

16.0, df = 1, P < 0.001).

Clark • BIRD REPELLENTS

o-Aminoacetophenone

Sham BONG Sham BONG

28 mM 14 mM

Treatment

Eig. 6. Average fluid intake for OAP as a function of a solution’s molar concentration,
surgery and treatment period. Starlings were presented with tap water during the pretreat-
ment period (hatched bars) and an OAP solution during the treatment period (solid bars).
Capped vertical bars depict one standard error.

DISCUSSION

When deprived of water during the day, starlings drank more water
than normal during the overnight recovery period. However, there was
no difference between pre- and post-test diurnal water consumption, ir-
respective of the experimental water deprivation schedule (Experiment 1 ).
Together these data suggest that the effects of short-term water depriva-
tion are ameliorated during the 18-h recovery period. This lack of car-
ryover, i.e., day effect, is consistent with previously reported results for
similar experiments (Clark and Shah 1991).

The use of the two Latinized square treatment presentation sequences
allowed a detailed analysis of the influence of delivery route on the avoid-
ance response (Experiment 2). Naive starlings that were exposed to only
the odor of OAP did not avoid the drinking apparatus, suggesting that
the odor of OAP was not repellent. Only when OAP was allowed to come
in contact with the mouth did starlings show an avoidance response.

Starlings failed to avoid the odor of OAP even after they had come in

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

■g

Methyl Anthranilate

Sham BONG
28 mM

Sham BONG
14 mM

Treatment

Lig. 7. Average fluid intake for MA as a function of a solution’s molar concentration,
surgery, and treatment period. Starlings were presented with tap water during the pretreat-
ment period (hatched bars) and a MA solution during the treatment period (solid bars).
Capped vertical bars depict one standard error.

oral contact with the substance (Experiment 2). Moreover, the prolonged
exposure to repellents (OAP and MA) over several days failed to result
in a learned odor avoidance (Experiment 3). These findings suggest that
starlings do not readily learn to avoid solutions associated with the odor
of repellents. The failure to form a conditioned avoidance is independent
of the history of exposure to odors or repellent solutions (Experiment 2).
Experiment 4 showed the lack of importance of olfactory cues in avoiding
repellents. Surgical elimination of the olfactory nerves did not cause birds
to increase consumption of repellent-bearing solution as was expected if
olfaction was the mediating mechanism for avoidance.

The lack of odor avoidance and contribution of the olfactory system
in the avoidance response is consistent with previous speculation that
sensory repellents are mediated by the trigeminal system, although the
role of taste cannot be ruled out on the basis of these experiments alone.
However, there is other evidence consistent with involvement of the tri-
geminal nerve for perception of OAP and MA. Mason et al. (1989)

Clark • BIRD REPELLENTS

showed that bilateral section of the OBTN virtually eliminated the avoid-
ance response of starlings to anthranilates. There is also electrophysio-
logical evidence that the OBTN is highly responsive to OAP and MA
(Clark, unpubl. data). Molecular modelling evidence suggests birds have
a receptor mechanism that is responsive to anthranilates and acetophen-
ones and is analogous to the vanalloid pain receptors in mammals (Clark
and Shah 1994). Finally, even after central taste nuclei are ablated, chick-
ens continue to avoid MA (Benowitz 1964), suggesting that taste is not
critical in mediating the avoidance response.

The unsuitability of oral exposure to repellents as the unconditional
stimulus may be attributable to its mode of action and/or localization. The
trigeminal repellents tested appear to be unpalatable but do not cause
illness (at least for the quantities voluntarily consumed by birds). In ad-
dition, the aversive effect is localized in the mouth (in this experiment)
and not in the gastrointestinal system. Thus, repellents that are unpalatable
and cause oral stimulation, in combination or alone, may be ineffective
as unconditional stimuli relative to repellents that cause illness and gas-
trointestinal stimulation. This does not suggest that trigeminal repellents
are poorer repellents than associative repellents. Both can be potent and
result in strong avoidance behavior. The difference between the two re-
pellent types is that birds fail to learn from their aversive experience with
trigeminal repellents.

From a practical viewpoint, failure to learn about the sensory attributes
of a repellent is not crippling to the objective of bird repellency. Labo-
ratory and field studies show birds continuously sample small quantities
of substances treated with sensory repellents (e.g.. Mason et al. 1985,
Clark and Shah 1991, 1994; Avery and Decker 1994). While total
amounts consumed are small, how does the trigeminal repellent work to
repel birds away from an area? The effectiveness of the repellent is two-
fold. In the first case, the repellent has a direct effect on individuals. When
movement is not restricted, a bird encountering a sensory repellent quick-
ly moves on to a more palatable resource patch (Mason et al. 1985). This
observation is consistent with optimal foraging theory (Charnov 1976)
and the proposition that low rates of energy return in a patch will favor
the probability of abandoning that patch (Lima 1985). The repellent can
be viewed as signalling low energy return or physically achieving the
fact. Regardless, the outcome is the same. Second, as a consequence of
reduced residency time in a patch, the recruitment opportunities to that
patch are reduced. Thus, the number of birds that might visit the patch
because they observe other individuals in the patch is reduced (Krebs
1974).

The inability to learn to stay away from sensory cues associated with

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

trigeminal repellents has implications for the evolution of plant-bird in-
teractions as well. The behavior for free-ranging birds is to sample fruit
throughout its development, apparently giving more attention to the sa-
lient chemical cues associated with palatability rather than ancillary sen-
sory cues (Willson and Comet 1993). To prevent untimely frugivory, it
is to the plant’s advantage to employ transient protection to fruits and
have birds return to the fruits at a later time when the seeds are ready for
dispersal, rather than have birds form a strong conditioned avoidance
response (e.g., Brower 1969, Guarino et al. 1974, Mason 1989). These
observations provide intriguing interpretive possibilities about how chem-
ical defenses of fruit should be structured to exploit the sensory systems
and learning capabilities of birds. Chemical defenses in unripe bird dis-
persed fruit should consist of trigeminal repellents and not compounds
that cause gastro-intestinal illness. Finally, a better understanding of how
plants prevent untimely frugivory will prove invaluable in the design of
nonlethal repellents for safe wildlife management strategies.

ACKNOWLEDGMENTS

I thank D. Coleman for assistance in the laboratory. C. A. Smeraski kindly performed
the histological preparations. This study was supported by the United States Dept, of Ag-
riculture cooperative agreement 12-34-41-0040 between the Monell Chemical Senses Center
and the Denver Wildlife Research Center. All experimental procedures meet guidelines set
forth by Monell’s institutional animal care committee.

LITERATURE CITED

Agree, T. E., E. H. Lavin, R. Nishida, and S. Watanabe. 1990. The serendipitous dis-
covery of ortho-aminoacetophenone as the foxy smelling component of labruscana
grapes. Chem. Eng. News 9:80.

Avery, M. L. and D. G. Decker. 1994. Responses of captive Fish Crows to eggs treated
with chemical repellents. J. Wildl. Manage. 58:261-266.

Benowitz, L. 1972. Effects of frebrain ablations on avoidance learning in chickens. Physiol.
Behav. 9:601-608.

Brower, L. P. 1969. Ecological chemistry. Sci. Am. 220:22-29.

Burd, G. D. 1993. Morphological study of the effects of intranasal zinc sulfate irrigation
on the mouse olfactory epithelium and olfactory bulb. Micros. Res. Tech. 24:195-213.
Charnov, E. L. 1976. Optimal foraging: the marginal value theorem. Theor. Popul. Biol.
9:129-136.

Clark, L. 1995. Modulation of avian responsiveness to chemical irritants: effects of pros-
taglandin El and analgesics. J. Exp. Zool. 271:432-4^40.

AND J. R. Mason. 1987. Olfactory discrimination of plant volatiles by the European

starling. Anim. Behav. 35:227-235.

AND . 1993. Interactions between sensory and postingestional repellents in

starlings: methyl anthranilate and sucrose. Ecol. Appl. 3:262-270.

AND P. S. Shah. 1991. Nonlethal bird repellents: in search of a general model

relating repellency and chemical structure. J. Wildl. Manage 55:538-545.

Clark • BIRD REPELLENTS

AND . 1993. Chemical bird repellents: possible use in cyanide ponds. J.

Wildl. Manage. 57:657-664.

AND . 1994. Tests and refinements of a general structure-activity model for

avian repellents. J. Chem. Ecol. 20:321—339.

, , and j. R. Mason. 1991. Chemical repellency in birds: relationship be-
tween chemical structure and avoidance response. J. Exp. Zool. 260:310-322.

AND C. A. Smeraski. 1990. Seasonal shifts in odor acuity by starlings. J. Exp. Zool.

177:673-680.

Crocker, D. R. and S. M. Perry. 1990. Plant chemistry and bird repellents. Ibis 132:300-
308.

Dolbeer, R. a. 1986. Current status and potential of nonlethal means of reducing bird
damage in agriculture. Pp. 474-483 in ACTA XIX Congresses Internationalis Orni-
thologici. Vol. 1 (H. Ouellet, ed.). Univ. of Ottawa Press, Ottawa, Ontario, Canada.

Garcia, .1. R., R. Kovner, and K. E Green. 1966. Cue properties vs. palatability of flavors
in avoidance learning. Psychonomic Sci. 20:313-314.

Getty, R. 1975. Sisson and Grossman’s The anatomy of the domestic animals. Vol. 2. Pp.
2022-2037. W. B. Saunders Company, Philadelphia, Pennsylvania.

Green, B. G., J. R. Mason, and M. R. Kare. 1990. Chemical senses, Vol. 2, Irritation.
Marcel Dekker, New York.

Guarino, j. L., W. E Shake, and E. W. Schafer. 1974. Reducing bird damage to ripening
cherries with methiocarb. J. Wildl. Manage. 38:338—342.

Krebs, J. R. 1974. Colonial nesting and social feeding as strategies for exploiting food
resources in the Great Blue Heron (Ardea herodias). Behaviour 51:99-134.

Lima, S. L. 1985. Sampling behavior of starlings foraging in simple patchy environments.
Behav. Ecol. Sociobiol. 16:135—142.

Mason, J. R. 1989. Avoidance of methiocarb-poisoned apples by red-winged blackbirds.
J. Wildl. Manage. 53:836-840.

AND L. Clark. 1992. Nonlethal repellents: the development of cost-effective, prac-
tical solutions to agricultural and industrial problems. Proc. Vert. Pest Conf. 15:115-
129.

AND R. E Reidinger. 1983. Importance of color for methiocarb-induced food aver-
sions in red-winged blackbirds (Agelaius phoencieus). J. Wildl. Manage. 47:383-393.

, M. A. Adams, and L. Clark. 1989. Anthranilate repellency to starlings: chemical

correlates and sensory perception. J. Wildl. Manage. 53:55-64.

, L. Clark, and P. S. Shah. 1991. Ortho-ami noacetophenone repellency to birds:

similarities to methyl anthranilate. J. Wildl. Manage. 55:334—340.

, J. F. Glahn, R. a. Dolbeer, and R. F. Reidinger. 1985. Field evaluation of

dimethyl anthranilate as a bird repellent livestock feed additive. J. Wildl. Manage. 49:
636-642.

Pelchat, M. L., H. j. Grill, P. Rozin, and J. Jacobs. 1983. Quality of acquired respon.ses
to tastes by Rattus norvegicus depends on type of associated discomfort. J. Comp.
Psychol. 97:140-153.

Schuler, W. 1983. Respon.ses to sugars and their behavioural mechanisms in the starling
{Stunms vulgaris L.). Behav. Ecol. Sociobiol. 13:243-251.

Statistica®. 1994. StatSoft, Inc., Tulsa, Oklahoma.

Walker, J. C., D. B. Walker, C. R. Tambiah, and K. S. Gilmore. 1986. Olfactory and
nonolfactory odor detection in pigeons: elucidation by a cardiac acceleration paradigm.
Physiol. Behav. 38:575—580.

Wenzel, B. M. and A. Salzman. 1968. Olfactory bulb ablation or nerve section and
behavior of pigeons in nonolfactory learning. Exper. Neurol. 22:472-^79.

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Willson, M. E and T A. Comet. 1993. Eood choices by Northwestern Crows: experiments
with captive, free-ranging and hand-raised birds. Condor 95:596—615.

Zahorik, D. M. 1976. Associative and non-associative factors in learned food preferences.
Pp. 181-200 in Learning mechanisms in food selection (L. M. Barker, M. R. Best, and
M. Domjan, eds.). Baylor Univ. Press, Waco, Texas.

Wilson Bull., 108(1), 1996, pp. 53-60

NEST-SITE SELECTION BY HOODED WARBLERS IN
BOTTOMLAND HARDWOODS OF SOUTH CAROLINA

John C. Kilgo, Robert A. Sargent,

Brian R. Chapman, and Karl V. Miller

Abstract. — We measured habitat features at 45 nests of Hooded Warblers (Wilsonia
citrina) and 45 non-use sites in bottomland hardwood habitats in the coastal plain of South
Carolina during the breeding seasons 1993—1994 to determine features that affect nesting
success. Hooded Warblers nested in switchcane {Arundinaria gigantea) and hardwood sap-
lings or shrubs that averaged 1.76 ± 0.10 m {SE) in height. Nests were more concealed
from above (P = 0.001) and from the side (P = 0.002) than surrogate nests placed at non-
use sites but were less concealed from below (P = 0.002). Nest sites also had a greater
number of potential substrates (P = 0.014) in the nest patch (5-m radius) and greater mea-
sures of vegetation density (P < 0.05) in the nest patch than non-use sites. Successful nests
differed from unsuccessful nests only in the amount of fern cover in the nest patch (greater
for successful nests, P = 0.012). Fern cover may influence nesting success through an effect
on behavioral defense strategies. Nesting success of Hooded Warblers may largely be un-
related to fine-scale differences in vegetative characteristics of the nest site. Received 28
Mar. 1995, accepted 20 Sept. 1995.

Because availability of suitable nest sites may be the most critical de-
terminant of habitat selection (and thus perceived habitat quality) by some
birds (Steele 1993), knowledge of what constitutes a suitable nest site, or
more importantly a successful nest site, is necessary (Martin 1993a). For
example, Martin and Roper (1988) found that successful Hermit Thrush
(Catharus guttatus) nests were characterized by a greater density of white
fir (Abies concolor) saplings in the 5-m radius circle surrounding the nest.
Such specific habitat features that affect nest fate should be identified for
other species.

Hooded Warblers (Wilsonia citrina) have been classified by the Part-
ners In Flight prioritization scheme as a species of “very high concern”
in the Southeast (Hunter et al. 1993a, b). We examined nest-site selection
patterns of Hooded Warblers to determine habitat differences between
successful and unsuccessful nests. We measured variables at two scales,
the nest site and the nest patch. Hooded Warblers inhabit moist mature
deciduous forests of eastern North America (Bent 1953, Powell and Rap-
pole 1986, Evans Ogden and Stutchbury 1994). In the coastal plain of
the southeastern United States, Hooded Warblers occur almost exclusively
in forested wetlands (Bent 1953) and reach their greatest abundance in
bottomland hardwood forests (Oak-Gum-Cypress [Quercus-Nyssa-Taxo-
dium] association).

Daniel B. Warnell School of Forest Resources, The Univ. of Georgia, Athens, Georgia 30602-2152.

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

STUDY AREA AND METHODS

We conducted this study at the U.S. Dept, of Energy’s Savannah River Site (SRS), a
78,000-ha tract in Aiken, Barnwell, and Allendale counties. South Carolina. These counties
lie in the Upper Coastal Plain physiographic province. Elevation ranges from <25 m at the
Savannah River to 80 m at headwater streams (Workman and McLeod 1990). Bottomland
hardwood forests are found along stream courses and may be seasonally flooded, usually
during late winter-early spring. Dominant canopy species include sweetgum {Liquidambar
styraciflua), swamp tupelo {Nyssa sylvatica var. biflora), red maple {Acer rubrum), water
oak {Quercus nigra), and diamond-leaf oak {Q. laurifolia). Dominants in the mid-story
include American holly, (Ilex opaca), sweet bay {Magnolia virginiana), red bay {Persea
borbonia), and ironwood {Carpinus Carolina). Switchcane {Arundinaria gigantea) and dog
hobble {Leucothoe axillaris) dominate the shrub layer, and ferns, primarily netted chain fern
{Woodwardia areolata) and Christmas fern {Polystichum acrostichoides), are the dominant
ground cover (Workman and McLeod 1990). Bottomland study sites ranged in width from
<50->1000 m and were adjacent to closed-canopy pine {Pinas elliottii and P. palustris)
forest.

We located Hooded Warbler nests in 1 1 bottomland hardwood strips during May-June
1993 and 1994 by observing adult behavior and by searching potential nesting habitat. We
found most nests during the incubation stage. We were unable to determine whether nests
were first or second attempts because individual birds were not marked and territories were
not mapped. We monitored nests at 3^ day intervals (Ralph et al. 1993) to determine nest
fate. Nests containing nestlings on the last visit before the expected fledging date were
assumed to have fledged. We defined successful nests as those that fledged at least one
nestling. Vegetation measurements were made following termination of the nesting attempt.
We made measurements at the nest plant and in the nest patch, defined as the 5-m radius
circle centered on the nest plant (Martin and Roper 1988). Vegetation measurements then
were repeated at an unused site. We located non-use sites by pacing 35 m (Ralph et al.
1993) upstream or downstream (determined by coin toss) in a direction parallel to the general
bearing of the bottomland strip. This procedure located non-use sites outside of the nest
patch but within the bottomland habitat. Non-use sites were centered on the plant stem
nearest to the 35-m point that was of the same species and approximate size as the substrate
plant (Ralph et al. 1993). Thus, equal numbers of nest sites and non-use sites were sampled.
Success data were obtained from 36 nests, 15 nests in 1993 (8 successful, 7 unsuccessful)
and from 21 nests in 1994 (10 successful, 1 1 unsuccessful). Eight additional nests in 1993
and one nest in 1994 that were empty when found were sampled and included in the
comparison of nest sites versus non-use sites but not in the analyses relating to nest success
(Martin and Roper 1988).

Measurements taken at the nest site included plant species used as the nesting substrate,
nest height, plant height, and percentage of nest concealment. Concealment indices (0-4: 0
= 0% concealed, 1 = 1-25% concealed, 2 = 26-50% concealed, etc.) were estimated by
viewing the nest from above and below and at nest level from a distance of 1 m in each of
the four cardinal directions (Martin and Roper 1988, Hoi way 1991 ). Lor concealment esti-
mates at non-use sites, an empty Hooded Warbler nest was placed at nest-height (i.e., the
height of the nest corresponding to the non-use site) in the suiTogate substrate plant (Holway
1991).

Measurements taken in the nest patch included overstory canopy cover, stem density of
potential nest substrates and trees, fern cover, other herbaceous ground cover, and vegetation
profile. Canopy cover above the patch was estimated by five hit-miss readings through an
ocular tube (James and Shugart 1970), one at the nest plant and one in each of the cardinal

Kilgo et al. • HOODED WARBLER NEST SITES

directions from the perimeter of the patch. Potential substrate and tree (woody stems > 3
m tall) densities were measured in five 1-m^ quadrats located randomly along the four
cardinal directions (transect and position on transect were randomized). Potential substrates
were defined as switchcane >1 m tall and other woody species 1—3 m tall. Percent foliage
cover of ferns and of other herbaceous ground cover also was estimated (0-4) within the
quadrats. Vegetation profile of the patch, which may be viewed as an index of concealment
at the scale of the patch, was determined using a 3-m tall vegetation profile board (Nudds
1977, Noon 1981) against which percentage cover was estimated (0^) for each 0.5-m
interval. The profile board was located at the nest plant and was read from a distance of 5
m in each of the cardinal directions.

For comparisons involving potential substrate density, nests were classified as either
“switchcane” or “other”, depending on the species of their actual substrate. Stem density
of switchcane then was determined for switchcane nests and of other for other nests. Thus,
the potential substrate variable was a nest-specific measurement which circumvented the
problem, e.g., of comparing average switchcane density across all nests when only a portion
of nests were in switchcane. Similarly, one vegetation profile measurement (nest level) was
nest-specific. The profile readings for the 0.5-m interval corresponding to the height of each
nest were compiled for the variable vegetation profile at nest level.

Univariate comparisons were made between Hooded Warbler nest sites and non-use sites
and between successful and unsuccessful nest sites for each habitat variable. Variables es-
timated with the 0^ index were compared with a Wilcoxon rank-sum test. All other com-
parisons were made with a two-sample f-test. Variances were assumed to be equal for
comparisons in which the sample sizes were the same. When sample sizes differed, the F-
test for equality of variance was used to test the equal variance assumption. Equal variance
tests always were appropriate. Because no differences (P > 0.05) were found between years
for any variables, data from both years were pooled.

RESULTS

Hooded Warblers selected saplings of nine different species as nest
substrates: switchcane, 20 (44%); red bay, 7 (16%); common gallberry,
5 (11%); American holly, 5 (11%); water oak, 2 (4%); diamond leaf oak,
2 (4%); blueberry (Vaccinium spp.), 2 (4%); wax myrtle {Myrica ceri-
fera), 1 (2%); and black oak (Quercus velutina), 1 (2%). Mean height of
the nest plant was 1.76 ± 0.10 m (SE). With one exception, in which the
nest was located in an upright branch of an American holly, nests were
placed in crotches of the main stem and primary branches of the substrate
plant. Nest height averaged 0.98 ± 0.36 m.

Hooded Warbler nest sites differed from non-use sites in several ways.
Concealment of nests from above and from the side was greater {P <
0.005) at nest sites than at non-use sites, but from below was lower (f
= 0.002) at nest sites than at non-use sites (Table 1). Potential substrate
density was greater {P = 0.014) at nest sites (3.79 ± 3.37 stems/m^) than
at non-use sites (2.08 ± 1.88 stems/m^). Vegetation profile measures for
all heights and at nest level were greater {P < 0.05) at nests sites than at
non-use sites (Table 1). Conversely, only one difference was determined

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Table 1

Comparison of Microhabitat Variables (x ± SE) at Hooded Warbler Nest sites (N
= 45) WITH Those at Random Sites (N = 45) within Bottomland Hardwoods,
Savannah River Site, South Carolina 1994-1994

Parameter

Nest site

Random site

Nest site

Nest height (m)

0.98

-h

0.05

1,01

-1-

0.05

Plant height (m)

1.76

-h

0.10

1.78

-h

0.14

Concealment

Side

1.93

0.10

1.50

0.08

0.002

Above

3.36

-i-

0.13

2.40

0.21

0.001

Below

1.58

-+-

0.16

2.09

0.14

0.002

Nest patch

Canopy coveC

4.36

-+-

0.10

4.49

0.10

0.360

Potential substrate density^

3.79

-+-

0.60

2.08

0.33

0.014

Pern cover“

1.19

-+-

0.12

1.31

0.16

0.842

Ground cover“

1.48

-+-

0.10

1.69

0.12

0.268

Vegetation profile'"

0.0— 0.5 m

3.07

-+-

0.10

2.68

0.13

0.025

0.5— 1.0 m

3.39

-+-

0.10

2.83

4“

0.14

0.000

1.0-1. 5 m

3.12

-H

0.09

2.36

0.14

0.000

1. 5-2.0 m

2.94

-+■

0.13

2.14

0.14

0.000

2.0-2.5 m

2.53

0.14

2.02

0.22

0.002

2. 5-3.0 m

2.30

-i-

0.14

1.79

0.13

0.007

Mean

2.88

-h

0.08

2.31

0.10

0.000

Nest level

3.47

-h

0.08

2.86

0.11

0.000

“ Index of percent coverage: 0 = 0%, 1

= 1-25%, 2 =

26-50%. 3

= 51-75%, 4

76-100%.

Index values were

compared with the Wilcoxon rank-sum test; all other comparisons were made with a two-sample r-test.

^ Estimated as the sum of five hit-miss readings taken within the patch (5 = total canopy closure).

Stem density (# stems/m^) of the plant species used as substrate. Substrates were categorized as switchcane or other
(woody stems 1. 0-3.0 m). Sample includes switchcane nests from 1993 and 1994 (N = 20) and other nests from 1994 (N
— 12; total sample = 32).

for the comparison between successful and unsuccessful nests: fern cover
was greater (P — 0.012) around successful nests.

DISCUSSION

Nest-site characteristics. — Switchcane best provides the structural fea-
tures of a nest substrate sought by Hooded Warblers of all plant species
occurring in bottomland hardwood habitats at SRS. It apparently is the
preferred substrate species throughout much of the southeastern United
States (Sprunt and Chamberlain 1949, Burleigh 1958). Switchcane is a
woody grass (Poaceae) that may grow to 10 m (Radford et al. 1964) but
normally ranges from 1-3 m. It commonly forms extensive thickets, or
canebrakes, in southeastern swamps. In addition to the nests that were in

Kilgo et al. • HOODED WARBLER NEST SITES

switchcane, many of the other nests were in saplings growing in cane-
brakes (i.e., switchcane provided most of the cover for most nests).

A variety of other plant species also were used as nest substrates. Hood-
ed Warblers reportedly prefer mountain laurel (Kalmia latifolia), Ameri-
can holly, and fetterbush {Lyonia spp.) in other parts of their range (Bent
1953). All of these species are thicket-forming shrubs (during the sapling
stage for American holly). Thus, Hooded Warblers may select shrubs,
regardless of species, not only for their microsite characteristics but also
for their thicket-forming properties, as evidenced by the greater density
of potential substrates within the nest patch than at non-use sites. Holway
(1991) found that species preference by another shrub-nesting warbler,
the Black-throated Blue Warbler (Dendroica caerulescens), also was site-
specific; they selected the understory shrub that offered the best protection
from weather and predators.

Hooded Warblers selected nest sites that were less concealed from be-
low than nests at non-use sites. Bent (1953, p. 613) quotes one author
who said that “the easiest way to locate a [Hooded Warbler] nest was to
place [his] head close to the ground, scan the low open spaces and look
for a clump of leaves, which sooner or later proved to be a nest.” The
adaptive advantage of an opening immediately below the nest is unclear,
though it may be related to escape strategies. Although Hooded Warblers
normally do not approach or leave the nest near the ground (Odum 1931),
when flushed, the female often drops from the nest straight to the ground
before flying away just above the ground for a short distance (J. C. Kilgo,
pers. obs.; Evans Ogden and Stutchbury 1994). Alternatively, such open-
ings may result simply from the greater shading provided by the under-
story.

Murphy (1983) and Martin (1992, 1993a) have suggested that preda-
tion, because it is the primary cause of nest failure, should be the key
factor influencing nest-site selection. Selection of nest sites with dense
vegetation theoretically can inhibit predator efficiency by visually screen-
ing the nest and parent activity, by providing too many potential nest sites
for the predator to search, and by physically impeding predators (Holway
1991). Our results indicate that Hooded Warblers may utilize each of these
strategies in their selection of nest sites. Hooded Warblers selected nest
sites that were better concealed from the side and from above than non-
use sites. Furthermore, nest patches contained a greater density of poten-
tial substrates and denser vegetation profiles at all heights than non-use
patches.

Effect of nest-site characteristics on success. — We detected no differ-
ence in concealment from any angle between successful and unsuccessful
nests. Similarly, Howlett and Stutchberry (in press) detected no effect of

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

nest concealment on predation of Hooded Warbler nests in Pennsylvania,
and Holway (1991) was unable to detect a relationship between conceal-
ment and the nest success of Black-throated Blue Warblers. We also failed
to detect a difference in the number of potential substrates between suc-
cessful and failed nests, as predicted by Martin and Roper (1988) and
Martin (1993a). Several factors may make statistical distinction of these
subtle habitat features difficult. First, Holway (1991) suggested that pred-
ators using olfactory cues would be less inhibited by visual concealment.
Furthermore, nest predation sometimes can be random, with some nests
being found by chance alone. Second, human visitation of nest sites dur-
ing monitoring may have increased the likelihood of predation, and thus
masked any effects of habitat on predation (Westmoreland and Best 1985,
Martin 1992). Finally, all nests of shrub-nesting woodland birds should
be concealed because selection of poorly concealed sites should be elim-
inated by natural selection (Wray and Whitmore 1979). The latter predic-
tion is contradicted by several authors who have detected differences in
concealment between successful and failed nests (e.g., Nolan 1978, Wray
and Whitmore 1979, Martin and Roper 1988). However, Wray and Whit-
more (1979), suggest that the apparently nonadaptive trait to select poorly
concealed nest sites may be maintained in Vesper Sparrows (Pooecetes
gramineus) because annual variation in their environment may permit the
occupancy of a variety of nest sites to be adaptive over time. Although
such temporal variation is probably great in the early successional habitats
of Vesper Sparrows, the environments of mature forests are relatively
stable. In addition, nest predation generally is higher in shrub and grass-
land habitats than in mature forests (Martin 1993b). Thus, a relationship
between concealment and success should not be as evident in forested
habitat. Studies of woodland shrub-nesting passerines support this con-
tention (Best and Stauffer 1980, Conner et al. 1986, Holway 1991, How-
lett and Stutchbury, in press; but see Martin and Roper 1988), whereas
results of studies of birds in earlier successional habitats are more variable
(Caccamise 1977, Best 1978, Nolan 1978, Wray and Whitmore 1979).
Much of the predation on shrub-nesting woodland birds may largely be
unrelated to fine-scale differences in concealment (Holway 1991).

The difference in fern cover between successful and failed nests is
intriguing. This finding may be related to nest-defense strategies. Female
Hooded Warblers almost invariably drop to the ground when flushed from
the nest, and rather than flying away, they often engage in a distraction
display, which consists of running through the underbrush with wings
drooped and tail spread (J. C. Kilgo, pers. obs.; Evans Ogden and Stutch-
bury 1994). This behavior likely is their primary (if not only) means of
nest defense. If insufficient ground cover exists in the patch to make this

Kilgo et al. • HOODED WARBLER NEST SITES

technique effective (i.e., if the bird must itself escape and is not able to
risk distracting the predator) the nest may be rendered more susceptible
to predation. Ferns may provide structure that conceals the displaying
female yet is sufficiently open to allow the predator to detect her. Thus,
degree of fern cover may be one of the subtle habitat features that deter-
mines nest fate of Hooded Warblers. This may also explain why the more
obvious measures of concealment and vegetation density did not differ
between successful and failed nests.

ACKNOWLEDGMENTS

This study was funded by the United States Dept, of Energy, Savannah River Site, the
United States Forest Service, Savannah River Forest Station Biodiversity Program, the Univ.
of Georgia, and Mclntire-Stennis Project No. GEO-0074-MS. J. Blake provided logistical
support. We thank the many field assistants who helped in locating and monitoring nests
and measuring vegetation. We thank G. W. Ware for providing statistical advice and A. S.
Johnson, K. C. Parker, R. J. Warren, and D. H. White for reviewing the manuscript. L. B.
Best and B. J. Stutchbury also provided valuable editorial comments.

LITERATURE CITED

Bent, A. C. 1953. Life histories of North American wood warblers. Bull. 203, U. S.
National Museum, Washington, D.C.

Best, L. B. 1978. Field Sparrow reproductive success and nesting ecology. Auk 95:9-22.
AND D. E Stauffer. 1980. Factors affecting nesting success in riparian bird com-
munities. Condor 82:149-158.

Burleigh, T. D. 1958. Georgia birds. Univ. of Oklahoma Press, Norman, Oklahoma.
Caccamise, D. E 1977. Nesting success and nest site characteristics in the Red-winged
Blackbird. Wilson Bull. 89:396—403.

Conner, R. N., M. E. Anderson, and J. G. Dickson. 1986. Relationships among temtory
size, habitat, song, and nesting success of Northern Cardinals. Auk 103:23-31.

Evans Ogden, L. J. and B. J. Stutchbury. 1994. Hooded Warbler {Wilsonia citrina). In
The birds of North America, No. 1 10 (A. Poole and F. Gill, eds.). The Acad. Nat. Sci.
of Philadelphia, Philadelphia, Pennsylvania; The American Ornithologists’ Union,
Washington, D.C.

Holway, D. a. 1991. Nest-site selection and the importance of nest concealment in the
Black-throated Blue Warbler. Condor 93:575-581.

Howlett, j. S. and B. j. Stutchbury. Nest concealment and predation in Hooded Warblers:

experimental removal of nest cover. Auk (in press).

Hunter, W. C., M. E Carter, D. N. Pashley, and K. Barker. 1993a. The Partners in
Flight prioritization scheme. Pp. 109-1 19 in Status and management of Neotropical
migratory birds (D. M. Finch and P. W. Stangel, eds.). U.S.D.A. For. Ser. Gen. Tech.
Rep. RM-229.

Hunter, W. C., D. N. Pashley, and R. E. F. Escano. 1993b. Neotropical migratory landbird
species and their habitats of special concern within the southeast region. Pp. 159-171
in Status and management of Neotropical migratory birds (D. M. Finch and P. W.
Stangel, eds.). U.S.D.A. For. Ser. Gen. Tech. Rep. RM-229.

James, F. C. and H. H. Shugart. 1970. A quantitative method of habitat description.
Audubon Field Notes 24:727—736.

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Martin, T. E. 1992. Breeding productivity considerations: what are the appropriate habitat
features for management? Pp. 455^73 in Ecology and conservation of Neotropical
migrant landbirds (J. M. Hagan and D. W. Johnston, eds.). Smithsonian Institution Press,
Washington, D.C.

. 1993a. Nest predation, nest sites, and birds: new perspectives on old patterns.

BioScience 43:523-532.

. 1993b. Nest predation among vegetation layers and habitat types: revising the

dogmas. Am. Nat. 141:897-913.

AND J. J. Roper. 1988. Nest predation and nest-site selection of a western population

of Hermit Thrush. Condor 90:51—57.

Murphy, M. T. 1983. Nest success and nesting habits of Eastern Kingbirds and other
flycatchers. Condor 85:208-219.

Nolan, V. 1978. The ecology and behavior of the Prairie Warbler Dendroica discolor.
Ornithol. Monogr. 26.

Noon, B. R. 1981. Techniques for sampling avian habitats. Pp. 42-51 in The use of mul-
tivariate statistics in studies of wildlife habitat (D. E. Capen, ed.). USDA Eor. Serv.
Gen. Tech. Rep. RM-87.

Nudds, T. D. 1977. Quantifying the vegetative structure of wildlife cover. Wildl. Soc. Bull.
5:113-117.

Odum, E. P. 1931. Notes on the nesting habits of the Hooded Warbler. Wilson Bull. 43:
316-317.

Powell, G. V. N. and J. H. Rappole. 1986. The Hooded Warbler. Pp. 827—853 in Audubon
wildlife report 1986 (R. L. Di Silvestro, ed.). National Audubon Society, New York,
New York.

Radford, A. E., H. E. Ahles, and C. R. Bell. 1964. Manual of the vascular flora of the
Carolinas. Univ. of North Carolina Press, Chapel Hill, North Carolina.

Ralph, C. J., G. R. Guepel, P. Pyle, T. E. Martin, and D. F. DeSante. 1993. Handbook
of field methods for monitoring landbirds. USDA For. Serv. Gen. Tech. Rep. PSW-144.

Sprunt, a. and E. B. Chamberlain. 1949. South Carolina bird life. Univ. of South Carolina
Press, Columbia, South Carolina.

Steele, B. B. 1993. Selection of foraging and nesting sites by Black-throated Blue War-
blers: their relative influence on habitat choice. Condor 95:568-579.

Westmoreland, D. and L. B. Best. 1985. The effect of disturbance on Mourning Dove
nesting success. Auk 102:774—780.

Workman, S. W. and K. W. McLeod. 1990. Vegetation of the Savannah River Site: major
community types. Publication SRO-NERP-19, National Environmental Research Park
Program. Savannah River Ecology Laboratory, Aiken, South Carolina.

Wray, T. and R. C. Whitmore. 1979. Effects of vegetation on nesting success of Vesper
Sparrows. Auk 96:802-805.

Wilson Bull., 108(1), 1996, pp. 61-71

CHANGE IN BODY MASS OF FEMALE COMMON
GOLDENEYES DURING NESTING AND
BROOD REARING

Michael C. Zicus' and Michael R. Riggs^

Abstract. — We measured body mass of female Common Goldeneyes {Bucephala clan-
giila) during nesting on fish bearing lakes in northcentral Minnesota, in 1982-1985. Median
body mass during egg laying was 775 g. Female mass during incubation varied among lakes
and possibly years. Mass at the start of incubation (698—715 g) was 10.7—11.0% greater
than that at hatching. Females regained most of the mass lost during incubation by the time
they abandoned their class IIC or class III ducklings. Goldeneyes in Minnesota weighed
less at the start of nesting than those studied on predominately fishless Ontario lakes; pro-
portional mass loss during incubation was also substantially less than that reported in Ontario
(approximately 20%). Differences in body mass dynamics may be related to the relative
ease of food acquisition during nesting; foods might be acquired more easily in more pro-
ductive wetlands despite the presence of fish. Received 24 Feb. 1995, accepted 1 Sept. 1995.

Relationships among incubation behavior, female body mass, and types
of nutrients and energy sources used by temperate nesting waterfowl are
understood reasonably well. In general, species that begin nesting earlier
have greater body mass, forage relatively less while nesting, rely more
on endogenous resources, and lose proportionately more mass during in-
cubation than do later nesting species (see review in Afton and Paulus
1992). Common Goldeneyes {Bucephala clangula) deviate somewhat
from this pattern. Although they are relatively small-bodied, females be-
gin nesting soon after arrival when many wetlands are still ice-covered.
Foraging territories also are defended vigorously during laying and early
incubation (Savard 1984, Zicus and Hennes 1993). In addition, laying
rates are low compared to other similar-sized waterfowl (cf Palmer 1976),
and clutch mass can exceed female mass (Zicus, unpubl. data). These
traits suggest that although females arrive with some stored reserves, ex-
ogenous nutrient sources may be important for clutch completion and
female maintenance during incubation.

Mallory and Weatherhead (1993) recently predicted that female Com-
mon Goldeneyes lose approximately 18.5% of their body mass during
incubation. Their prediction was based on relationships proposed by Af-
ton and Paulus ( 1 992) and appeared to be supported by data from an
Ontario study where wetlands had been influenced extensively by acid

‘ Minnesota Dept, of Natural Resources, Wetland Wildlife Populations and Research Group, 102 2.3rd
St., Bemidji, Minnesota 56601.

^ Minnesota Dept, of Natural Resources, Wildlife Populations and Research Unit, 500 Lafayette Road,
St. Paul, Minnesota 55155.

THE WILSON BULLETIN • Vol. 108, No. I, March 1996

deposition (Mallory et al. 1994). Harvey et al. (1989) concluded that
Wood Duck {Aix sponsa) body mass dynamics during incubation varied
substantially among and within individuals. They speculated that vari-
ability in Wood Duck incubation mass was related in part to fluctuating
environmental conditions. Unfortunately, few published data relate local
or regional environmental factors to body mass changes in other species
of incubating waterfowl.

We describe body mass changes during nesting and brood rearing for
female Common Goldeneyes in Minnesota. We examined the effects of
different lakes and years and of changing reproductive stage on female
body mass. Goldeneyes have been studied previously in areas where fish
were absent or where goldeneyes appeared to favor fishless wetlands
thereby avoiding dietary competition with fish during the reproductive
period (Eriksson 1979, Eadie and Keast 1982, Blancher et al. 1992, and
others). Whereas many lakes in Mallory and Weatherhead’s 1993 study
area were fishless (Mallory et al. 1994), each of our study lakes supported
fish communities. Thus, our data should improve the understanding of
goldeneye nesting biology where they commonly occupy fish-bearing
lakes.

STUDY AREA AND METHODS

Lemale Common Goldeneyes were weighed during nesting on three lakes in Beltrami and
Itasca counties of northcentral Minnesota. Refuge Pond, North Twin Lake, and Island Lake
differed in size (46, 117, and 1250 ha, respectively), amount and type of public use, and
use by goldeneyes (Zicus et al. 1995). The two larger lakes supported fish populations
(dominated by Centrarchidae, Percidae and Esocidae) and were characterized by morpho-
edaphic indices (MEI) (Ryder 1965) of 6.75 and 18.01, respectively (Minn. Dept. Nat.
Resour., Section of Lisheries, unpubl. data). These values are near optimum for highly
productive fish communities (Ryder et al. 1974). The smallest lake supported only minnows.
Lemales with hatched young were captured and weighed on Refuge Pond and Island and
North Twin Lakes as well as on 10 additional lakes, each of which supported productive
fish populations with species composition similar to those in the lakes where nesting ducks
were weighed. These additional lakes had MEIs ranging from 4.52 to 21.78 (Minn. Dept.
Nat. Resour., Section of Lisheries, unpubl. data).

Lemale Common Goldeneyes were captured before incubation began with nest traps (Zi-
cus 1989) and again when we inspected nest boxes for use. Lemales were leg banded with
U.S. Lish and Wildlife Service bands. Incubating female mass was measured when possible
during weekly nest checks, and females accompanying broods were weighed when they
were caught nest prospecting (Zicus and Hennes 1989) and during annual leg banding
(Johnson 1972). Mass was determined to the nearest 5 g using spring scales and was un-
adjusted for female structural size.

Reproductive stage for egg-laying females was defined relative to the start of incubation,
and that of incubating and brood-rearing females was referenced to the departure of young
from nests. This differs from the convention often used for incubating females. However,
we believe it is preferable because it allows corresponding days to be compared more
appropriately. Reproductive stage of females with young from unmonitored nests was esti-

Zicus and Riggs • GOLDENEYE BODY MASS

mated from the age of the majority of the ducklings in the brood. Duckling age was deter-
mined by comparison with known-age ducklings in various stages of plumage development.
Plumage stages were assigned the following ages: IB — 10 days, IC — 18 days, II A 27 days,
IIB — 35 days, and IIC — 44 days.

We examined mass change of females before incubation using linear regression (PROC
GEM; SAS Institute Inc. 1988). Year and location effects could not be examined because
too few females were captured. Most females were weighed more than once during incu-
bation, so we investigated their mass change using a generalized linear mixed model
(GLMM) with maximum likelihood estimators (PROC MIXED; SAS Institute Inc. 1992).
This approach allows measurements on subjects to be repeated within and across years.
Dependencies among repeated measures are modelled explicitly and ensuing tests are ad-
justed for this dependence based on the underlying covariance structure (Laird and Ware
1982, Ware 1985). We determined (Jennrich and Schluchter 1986) that a compound sym-
metry covariance structure was optimal for our models. We modelled the effect of lake,
year, linear, quadratic, and cubic effects of incubation day, and their interactions on female
mass. When interactions were not significant (a = 0.05), we used a reduced model. Log-
likelihood ratio statistics were used to evaluate model goodness of fit, and simultaneous
paired comparisons were made using a Bonferroni adjustment to pairwise differences in the
time-adjusted means (Dobson 1990). Brood-rearing females were measured only once and
their mass change was examined using linear regression (PROC GLM; SAS Institute Inc.
1988). We ignored possible lake and year effects because too few brood-rearing females
were measured.

We further examined nonsignificant statistical results using post hoc power analyses
(Anonymous 1995). Regression results were evaluated using a SAS MACRO (Latour 1992).
Power calculations for the generalized linear mixed model were based on adjusted least
squares effects estimated by PROC MIXED (SAS Institute Inc. 1992).

RESULTS

From 1982 to 1985, 45 females were weighed prior to or during egg-
laying. In addition, 82 females were weighed repeatedly (1-5 times each
year) for a total of 213 times at known points during incubation or with
hatched young in the nest. One female was weighed in four years, five
in three years, 25 in two years, and 51 in only one year for a total of 120
within-year time-series. During brood rearing, 63 females were weighed.

Prelaying and laying. — Reproductive status of females weighed before
the start of incubation varied (Table 1). Those considered known nesters
successfully incubated nests that we observed. Known nesters were
weighed from one to 30 days before incubation (median = 15.5 days)
and most likely represented females that were beginning to lay the clutch
that they eventually incubated. In contrast, the sample of unknown status
likely included females nesting elsewhere as well as those laying eggs
parasitically when captured. We could not detect any linear trend in mass
of known nesters during the laying period (mass change = 1.5 g/day,
95% confidence interval = -0.4 to 3.5).

Incubation. — We fit the GLMM to measurements of mass for females
that successfully incubated a clutch. We detected no significant interac-

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Table 1

Mass of Eemale Common Goldeneyes Captured During the Laying Period in
Northcentral Minnesota, 1982-85

Mass (g)

Stage (daysp

Status

Median

Range

Median

Known nesters

768

34.0

775

685-810

15.5

Unknown

733

43.8

720

660-830

* Days prior to the start of incubation.

tions among the main effects (lake and year) and linear, quadratic, or
cubic measures of incubation day (all Ps > 0.46) that affected mass loss.
Two models, one linear and one cubic with incubation day, indicated mass
was influenced by stage of incubation. The cubic model fit the data best
(G“ = 16.48, df = 2, P < 0.001) and indicated a curvilinear effect of
stage on female mass (Table 2 and Fig. 1).

Body mass also was influenced by conditions related to specific lakes
(P = 2.72; df = 2, 79; P = 0.072). Female body mass was greater
throughout incubation on Island Lake than on North Twin Lake (P =
4.29; df = 1, 79; P = 0.041), but mass did not differ between Refuge
Pond and either North Twin (P = 0.00; df = 1, 79; P = 0.980) or Island
Lake (P - 1.94; df = 1, 79; P = 0.167) (Fig. 1).

We detected no effect due to different years with the GLMM (P =
0.75; df = 3, 125; P = 0.523), but graphical comparison of body mass
of females for whom we had measurements in 1982 and 1983 suggested
year might have an effect. Each of the four females was weighed at
similar points during incubation both years, and each was 20-^0 g lighter
in 1983 than 1982 (Fig. 2).

Four of eight females that abandoned their nests during incubation had

Table 2

Estimated Coefficients for Polynomial Measures of Incubation Day on Mass
Change in Eemale Common Goldeneyes from Three Northcental Minnesota Lakes,

1982-1985

Measure

SE (P)

Day

6.3027

1.1799

<0.001

Day^

-0.4127

0.0995

<0.001

Day^

0.0083

0.0023

<0.001

“Two-sided test 3 = 0.

Zicus and Riggs • GOLDENEYE BODY MASS

Fig. 1. Maximum likelihood estimates of cubic temporal trend in female Common Gold-
eneye body mass during incubation on three northcentral Minnesota lakes, 1982-1985.

a body mass greater than that predicted for successful females at the
comparable point in incubation, and four were less than predicted.

Brood rearing. — Females in the composite sample regained body mass
at approximately 2 grams/day (F = 51.7; df = 1, 62; P < 0.001) from
the low they had reached when the young departed the nest (Fig. 3).
Females apparently regained most of the body mass lost during incubation
by the time they left their broods of class IIC or class III ducklings to
molt.

DISCUSSION

Our results differed markedly from previous measurements of Common
Goldeneye mass. Minnesota goldeneyes appear to weigh less at the onset
of nesting than do Ontario birds. Mallory (1991:17) reported that prelay-
ing females averaged 875 g at his Wanapitei study site and 842 g at a
site farther east. These values are 67-100 g more than our median at the
start of laying and 12^5 g more than our heaviest females. Whether
females truly differ to this extent is unclear. Our sample of prelaying
females was small, and we weighed females at various times during pre-
laying and laying. Furthermore, the Wanapitei values are estimates ob-
tained by adjusting female mass determined during incubation to account

THE WILSON BULLETIN • VoL 108, No. 1, March 1996

Days Before Departure From Nest

Fig. 2. Body mass of individual female Common Goldeneyes measured during incu-
bation while nesting on three northcentral Minnesota lakes, 1982 and 1983. Individual fe-
males are identified by unique symbols.

for incubation mass loss and that assumed lost in the course of laying a
clutch (Mallory 1991:16-17). Goldeneye females have been reported to
lose approximately 22 g/egg (i.e., approximately 1 1 g/day) during egg
laying (H. G. Lumsden, unpubl. data cited in Mallory 1991). We did not
detect this sort of change in Minnesota, and there was sufficient power
(>0.99) to detect a change of as little as 6 g/day. Projected mass of
Wanapitei females may be biased high because mass loss per egg in
Ontario is now believed to be <22 g (M. L. Mallory, Environ. Can., pers.
comm.). Nonetheless, model predictions of mass for Minnesota females
at the start of incubation ranged from 698-715 g depending on factors
associated with the nesting lake. In contrast, Ontario goldeneyes appear

ZicLis and Riggs • GOLDENEYE BODY MASS

Fig. 3. Linear regression predictions (±95% confidence limits) and observed mean body
mass (sample sizes) for female Common Goldeneyes with hatched young on 12 northcentral
Minnesota lakes, 1982—1985.

to begin incubation at 752-829 g (calculated from Mallory and Weath-
erhead 1993).

Minnesota goldeneye females lost proportionately less mass during in-
cubation than Ontario birds. Rate of mass change did not differ among
nesting lakes or years, although our analysis indicated location and pos-
sibly year affected overall incubation mass. The proportionate mass loss
by goldeneyes from the start of incubation until broods departed the nest
(31 days) was 10.7-11.0% depending on the nesting lake. In comparison,
Mallory and Weatherhead (1993) estimated mean mass loss variously and
reported changes for Ontario goldeneyes of 16.7% (page 853), 17.8%
(page 856), and 24.5% (calculated from equation page 853 using 31 days
of incubation).

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Incubating Common Goldeneyes apparently lose body mass in a non-
linear fashion (Mallory and Weatherhead 1993, this study). However, our
analysis suggested a different pattern than that described for birds in On-
tario. Minnesota goldeneyes experienced a rapid initial decline during
early incubation, reduced rate of loss in mid-incubation, and an increased
rate of loss again prior to hatching and duckling departure. Mass loss
during incubation for Ontario females has been modelled as decreasing
monotonically with the lowest mass occurring on the last day of incu-
bation (Mallory and Weatherhead 1993:equation p. 853). However, Mal-
lory and Weatherhead (1993:853) also indicated that some females
reached their lowest mass as early as day 18 of incubation, after which
some mass was regained.

Female mass when broods departed the nest was similar in Ontario and
Minnesota, but again measurements are not directly comparable. Mallory
and Weatherhead (1993:856) reported that mean body mass of 10 females
at the end of incubation was 626 g. However, two of the females included
in their sample had deserted their nests after 18 and 24 days of incubation,
respectively. By comparison, Minnesota females were predicted to weigh
615-633 g on the departure day (model intercepts) depending on the lake.
Unadjusted arithmetic mean mass on the day broods departed the nest
was 616 g (N = 20).

Mallory and Weatherhead (1993) speculated that female goldeneyes
that lost too large a proportion of their initial incubation weight or
dropped below approximately 600 g might be more prone to nest aban-
donment than others. We observed no indication that mass of females
deserting their nests differed from that of successful females, but more
data are needed. Furthermore, seven of 20 successful females weighed
the day ducklings departed the nest were less than 600 g and five were
<580 g. Some females remained below 600 g well into brood rearing.

Methodology alone does not explain the marked differences in incu-
bation mass and proportional mass change during incubation in our study
versus those reported for Ontario goldeneyes. At least two explanations
are tenable. In addition to the well known difference existing in geese,
body size has been shown to vary geographically in some passerines
(Aldrich and James 1991, Twedt et al. 1994). Ontario females may be
structurally larger than those nesting in Minnesota and thus able to return
to breeding areas with more stored reserves. Alternatively, Ontario fe-
males may be similar in size but may return with proportionately more
stored resources (i.e., better condition). Several studies (e.g., Gatti 1983,
Harvey et al. 1989, Aldrich and Raveling 1983) have reported that heavier
individuals lost a greater proportion of their body mass in incubation than
lighter conspecifics. Gatti (1983) reasoned that heavier Mallards {Anas

Zicus and Riggs • GOLDENEYE BODY MASS

platyrhynchos) could afford to lose more mass than those in poorer con-
dition. In contrast, Kennamer and Hepp (1987) reported that double-
brooded Wood Duck females lost a smaller proportion of their body mass
after their first nesting than single-brooded females and may have been
in better condition as a result.

Whether Ontario females are structurally larger than those in Minnesota
or just begin incubation in better condition, incubation constancy was
similar in the two locations (Mallory and Weatherhead 1993, Zicus et al.
1995). Together with the disparate mass loss and comparable weights at
the end of incubation, these results indicate that either differences in. for-
aging time during incubation recesses exist or else Ontario females con-
sume less food or food of lower quality during incubation than do Min-
nesota females. Most resident nesting goldeneyes from Island Lake and
Refuge Pond appeared to forage exclusively on their respective lakes,
whereas most females nesting on North Twin Lake foraged elsewhere.
Zicus and Hennes (1993) observed nesting female goldeneyes feeding at
least as much as most small-bodied waterfowl which rely extensively on
exogenous resources. They also reported that time devoted to foraging
during nesting varied among years and concluded that females foraged
less when food was most available. Harvey et al. (1989) likewise believed
that reduced food availability in some years contributed to a greater rel-
ative mass loss during incubation in Wood Ducks. Furthermore, incuba-
tion mass was lowest on Refuge Pond, the location among our three study
sites where Zicus et al. (1995) reported low incubation constancy in a
concurrent study. They speculated that low constancy was a consequence
of increased foraging time because of more difficult food acquisition.

Goldeneye mass during incubation varied among the lakes that we stud-
ied. Although we detected no differences among years with the GLMM,
measurement of a small sample of the same females in two consecutive
years suggested that yearly differences of at least 20 g might exist in
some years. The GLMM analysis had low power (0.46) at a = 0.05 to
detect such a difference. Nonetheless, the among lake and year mass
differences we measured were less than differences between Minnesota
and Ontario. Mann and Sedinger (1993) suggested that Northern Pintail
{Anas acuta) females nesting in Alaska relied more on endogenous re-
sources than temperate nesting congeners because of less productive high
latitude wetlands. Goldeneyes might use different nesting strategies de-
pending on average environmental conditions and food availability in the
regions they occupy. In some regions, wetland productivity may be suf-
ficient to provide adequate goldeneye food availability even in the pres-
ence of fish (DesGranges and Gagnon 1994:220). Resource acquisition
would then be less constrained by female foraging time because food

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

could be acquired easily. Patterns of body mass change before and during
incubation indicate this situation is likely in Minnesota. Alternatively,
females might rely more on resources acquired before arrival to nesting
areas where wetland productivity is low. If nesting strategy within the
species is flexible, relationships among habitat quality, female mass, and
reproductive effort and success (Mallory et al. 1994) may need to be
reexamined.

ACKNOWLEDGMENTS

We thank N. L. Weiland, biologist on the Chippewa National Lorest Blackduck District,
and private landowners on North Twin and Island Lakes for permission to work with their
waterfowl nest boxes. Numerous summer technicians with the Minnesota Department of
Natural Resources helped capture and weigh brood females during routine leg banding.
Discussions with and comments by M. A. Hanson, S. J. Maxson, R. T. Eberhardt, and D.
R Rave improved the manuscript. M. L. Mallory also reviewed an earlier draft of the
manuscript.

LITERATURE CITED

Afton, A. D. AND S. L. Paulus. 1992. Incubation and brood care, Pp. 62-108 in Ecology
and management of breeding waterfowl (B. D. J. Batt, A. D. Afton, M. G. Anderson,
C. D. Ankney, D. H. Johnson, J. A. Kadlec, and G. L. Krapu, eds.). Univ. of Minnesota
Press, Minneapolis, Minnesota.

Aldrich, J. W. and E C. James. 1991. Ecogeographic variation in the American Robin
(Turdus migratorius). Auk 108:230—249.

Aldrich, T. W. and D. G. Raveling. 1983. Effects of experience and body weight on
incubation behavior of Canada Geese. Auk 100:670—679.

Anonymous. 1995. Journal news. J. Wildl. Manage. 59:196-198.

Blancher, P. j., D. K. McNicol, R. K. Ross, C. H. R. Wedeles, and P. Morrison. 1992.
Towards a model of acidification effects on waterfowl in Eastern Canada. Environ.
Pollut. 78:57-63.

Desgranges, J-L. and C. Gagnon. 1994. Duckling response to changes in the trophic web
of acidified lakes. Hydrobiologia 279/280:207—220.

Dobson, A. J. 1990. An introduction to generalized linear models. Chapman and Hall.
London, England.

Eadie, j. M. and a. Keast. 1982. Do goldeneyes and perch compete for food? Oecologia
55:225-230.

Eriksson, M. O. G. 1979. Competition between freshwater fish and goldeneyes Bucephala
clangula (L.) for common prey. Oecologia 41:99-107.

Gatti, R. C. 1983. Incubation weight loss in the Mallard. Can. J. Zool. 61:565-569.

Harvey, W. E, IV, Hepp, G. R., and R. A. Kennamer. 1989. Body mass dynamics of wood
ducks during incubation: individual variation. Can. J. Zool. 67:570—574.

Jennrich, R. I. AND M. D. SCHLUCHTER. 1986. Unbalanced repeated-measures models with
structured covariance matrices. Biometrics 42:805—820.

Johnson, L. L. 1972. An improved capture technique for flightless young goldeneyes. J.
Wildl. Manage. 36:1277-1279.

Kennamer, R. A. and G. R. Hepp. 1987. Frequency and timing of second broods in Wood
Ducks. Wilson Bull. 99:655-662.

Zicus and Riggs • GOLDENEYE BODY MASS

Laird, N. M. and J. H. Ware. 1982. Random-effects models for longitudinal studies.
Biometrics 38:963-974.

Latour, K. R. 1992. %Power: a simple macro for power and sample size calculations.
Proc. SAS Lfser’s Group Int. 17:1173—1177.

Mallory, M. L. 1991. Acid precipitation, female quality, and parental investment of Com-
mon Goldeneyes. M.S. thesis, Carleton Univ., Ottawa, Ontario.

AND P. J. Weatherhead. 1993. Incubation rhythms and mass loss of Common

Goldeneyes. Condor 95:849-859.

, D. K. McNicol, and P. j. Weatherhead. 1994. Habitat quality and reproductive

effort of Common Goldeneyes nesting near Sudbury, Canada. J. Wildl. Manage. 58:
552-560.

Mann, E E. and J. S. Sedinger. 1993. Nutrient-reserve dynamics and control of clutch
size in Northern Pintails breeding in Alaska. Auk 110:264—278.

Palmer, R. S. 1976. Handbook of North American birds, Vol. 3. Yale Univ. Press, New
Haven, Connecticut.

Ryder, R. A. 1965. A method for estimating the potential fish production of north-tem-
perate lakes. Trans. Amer. Fish. Soc. 94:214—218.

, S. R. Kerr, K. H. Loftus, and H. A. Regier. 1974. The morphoedaphic index, a

fish yield estimator — review and evaluation. J. Fish. Res. Board Can. 31:663—688.

SAS Institute Inc. 1988. SAS/STAT user’s guide, release 6.03 edition. SAS Institute Inc.,
Cary, North Carolina.

. 1992. SAS technical report p-229: SAS/STAT software changes and enhancements,

release 6.07. SAS Institute Inc., Cary, North Carolina.

Savard, J.-P. L. 1984. Territorial behavior of Common Goldeneye, Barrow’s Goldeneye
and Bufflehead in areas of sympatry. Ornis Scand. 15:211-216.

Twedt, D. j., W. j. Bleier, and G. M. Linz. 1994. Geographic variation in Yellow-headed
Blackbirds from the northern Great Plains. Condor 96:1030—1036.

Ware, J. H. 1985. Linear models for the analysis of longitudinal studies. Am. Stat. 39:95-

101.

Zicus, M. C. 1989. Automatic trap for waterfowl using nest boxes. J. Field Ornithol. 60:
109-111.

AND S. K. Hennes. 1989. Nest prospecting by Common Goldeneyes. Condor 91:

807-812.

AND . 1993. Diurnal time budgets of breeding Common Goldeneyes. Wilson

Bull. 105:680-685.

, , AND M. R. Riggs. 1995. Common Goldeneye nest attendance patterns.

Condor 97:461^72.

Wilson Bull., 108(1), 1996, pp. 72-79

INTERSPECIFIC VARIATION IN THE CALLS OF
SPHENISCUS PENGUINS

Nina N. Thumser,' - Jeffrey D. Karron,' and
Ml LUCENT S. FiCKEN'

Abstract. — We compared the vocal repertoires of Jackass (Spheniscus demersus), Hum-
boldt (5. humboldti), and Magellanic (5. rnagellanicus) penguins. Discriminant and cluster
analyses of the bray call indicate that Jackass and Magellanic penguins are more similar to
each other than either is to the Humboldt penguin, and all three are distinct from the Rock-
hopper penguin (Eudyptes chrysocome). The congruence of the vocal analyses with phylog-
enies based on allozyme data suggests that differences in vocalizations reflect gradual di-
vergence over time, not character displacement. Received 1 Jan. 1995, accepted 20 Sept.
1995.

Vocalizations frequently are used to assess taxonomic relationships in
birds and the use of song in avian systematics has been thoroughly re-
viewed by Payne (1986). Vocalizations also have been used to determine
phytogenies of non-passerine species such as the caledrine sandpipers
(Miller et al. 1988). Jouventin (1982) found calls to be the best behavioral
criterion for classifying penguins. The degree of variation in calls was
used to infer subspecies and species status in island populations of nine
penguin taxa. Vocalizations have been shown to be of primary importance
in the communication of many penguin species (Pettingill 1960, Stone-
house 1960, Boersma 1974, Spurr 1975, Jouventin 1982). Although be-
havior has been studied in all of the Spheniscus penguins, only prelimi-
nary information exists concerning their vocalizations (Boersma 1974,
1976, Eggleton and Siegfried 1979, Jouventin 1982, Scolaro 1987).

Species status and phylogenetic relationships in the genus Spheniscus
are not clearly defined. There is insufficient detail in the fossil record to
distinguish among species (Simpson 1976). Morphological studies led
Clancey (1966) to classify Jackass penguins (5. demersus) as a subspecies
of Magellanic penguins {S. rnagellanicus). O’Hare (1989) used 22 mor-
phological characters to clearly differentiate Spheniscus from other pen-
guin genera. However, he was unable to determine the taxonomic rela-
tionships among species within the genus. Utilizing data from DNA-DNA
hybridization, Sibley and Monroe (1990) proposed that S. demersus be
viewed as a superspecies containing demersus, rnagellanicus, and the
Humboldt penguin (S. humboldti). Recent allozyme analyses suggest that

' Dept, of Biological Sciences, P.O. Box 413, Univ. of Wisconsin-Milwaukee, Milwaukee, Wisconsin
53201.

^ Present Address: Dept, of Biological Sciences. 150 W. University Blvd. Florida Tech. Melbourne, Flor-
ida 32901-6988.

Thumser et al. • SPHENISCUS VOCAL COMPARISON

Humboldt penguins form a distinct species, but that Jackass and Magel-
lanic penguins are closely related (Grant et al. 1994, Thumser and Karron
1994).

We quantitatively analyzed the vocalizations of three Spheniscus spe-
cies (Jackass, Humboldt, and Magellanic penguins) and one outgroup
(Rockhopper penguins, Eudyptes chrysocome). Their vocal repertoires
were compared to determine if species consistently differ in the acoustical
structure of their calls. A resulting phylogeny was compared to an inde-
pendent phylogeny based on protein polymorphisms (Thumser and Kar-
ron 1994).

METHODS

Vocalizations of 21 Humboldt penguins were recorded at the Milwaukee County Zoo in
Wisconsin (February 1986— May 1987, February— March 1988), the Brookfield Zoo in Chi-
cago, Illinois (November 1987- April 1988), and the St. Louis Zoo in Missouri (October
1988). Recordings of 12 Jackass penguins were made at the Henry Villas Zoo in Madison,
Wisconsin (January-April 1988), the Knoxville Zoo in Tennessee (May 1989), and the
Racine Zoo in Wisconsin (February-March 1990). Seven Magellanic penguins were record-
ed at the Cincinnati Zoo in Ohio (April 1988) and by Jim Klinesteker at the John Ball
Zoological Gardens in Grand Rapids, Michigan (Spring 1989). Eleven Rockhopper penguins
were recorded at the Cincinnati Zoo (April 1988) and the St. Louis Zoo (October 1988).

This study was performed exclusively on captive penguins. Although the majority of
recorded Jackass penguins were born in captivity, most of the Magellanic, Rockhopper, and
Humboldt penguins were born in the wild. The birds comprising these captive populations
may have been drawn from a limited number of wild populations. However, the results from
this study should be representative since seabirds usually have limited variation in their
vocalizations, particularly at or below the species level (Pierotti 1987).

Observations were made during breeding periods, mainly prior to and just after egg laying,
since most of the calls occurred at those times. A microphone was placed inside the exhibit,
but observations were made from outside tbe exhibit to minimize disturbance of the birds.
Individuals were identified by tag color. General behavior was observed throughout the day
(from 08:00 to 17:00 h CST). Notes were recorded on videotape (Hitachi HJ 5000) and by
hand. Vocalizations were recorded throughout the period on a cassette recorder (Aiwa HSJ
500) using a Nakamichi (CM 100) microphone. Whenever possible the identity of the caller
was noted. Peak periods of vocalization were simultaneously videotaped and tape-recorded.

The recorded calls were analyzed at 150 Hz bandwidth using a Kay 7800 Digital Sona-
Graph and digitized using a Sigma Scan (1988) Program. For the bray call, the number of
syllables per call, total duration of the call, sum of the inter-syllable intervals, duration of
the longest syllable, and minimum, main, and maximum frequency of the longest syllable
were recorded. The main frequency represented the darkest band in the sonagram of the
call. These seven variables were selected to assess the acoustic structure of the bray call
based on both frequency and temporal components. In addition, these parameters were
selected because they could be measured precisely.

The vocalizations were analyzed using discriminant and cluster analysis in SYSTAT (Wil-
kinson 1990). The bray call was selected for analysis since nested ANOVA of individuals
within populations within species indicated significant differences at the species level for
more than one parameter (Thumser 1993). The data set included all recorded bray calls of
the three Spheniscus penguins and the Rockhopper penguins. There were 109 calls from

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Table 1

Discriminant Eunction 1 Showing the Correlations between Conditional Dependent
Variables and Dependent Canonical Eactors in Eour Penguin Species

Factor

Variable

Number of syllables

-0.289

-0.710

-0.024

Total duration

0.337

-0.652

0.401

Inter-syllable interval

0.270

0.018

-0.345

Duration of longest syllable

0.622

-0.229

0.559

Maximum frequency

0.008

-0.635

-0.075

Minimum frequency

-0.145

0.464

0.083

Main frequency

-0.292

0.257

0.544

Chi-square

496.75

213.17

39.61

<0.001

Correlation

0.764

0.644

0.339

Humboldt, 77 calls from Jackass, 38 calls from Magellanic, and 106 calls from Rockhopper
penguins. Each of the bray call variables was standardized by converting its values to z-
scores prior to analysis. In discriminate analysis known groups were used to generate linear
models which gave the best fit for that grouping. The data were also analyzed to determine
how well the model predicts the actual groupings. Another multivariate technique, cluster
analysis, was used to detect natural groupings in data with no prior expectations. In this
case, Pearson’s distance measures and the single-linkage method were performed by cal-
culating the mean of each of the standardized variables for each species.

RESULTS

For the first discriminant function. Factor 1 arranged the four species
primarily on the basis of the duration of longest syllable and the total
duration of the call, while Factor 2 was primarily based on the number
of syllables, the total duration of the call, and the maximum frequency
of the longest syllable (Table 1). Overall, the analysis correctly catego-
rized 86% of Humboldt, 82% of Jackass, 52% of Magellanic, and 79%
of Rockhopper penguin calls. These vocal parameters clearly separated
the Spheniscus penguins from the outgroup, Rockhopper penguins (Fig.
1 A). Therefore, the outgroup was removed from the analysis and a second
discriminant analysis was run to increase the spread among the Spheniscus
penguins. In this discriminant function. Factor A arranged the three spe-
cies primarily on the basis of syllable number and maximum frequency
of the longest syllable, and Factor B was based primarily on the duration
and main frequency of the longest syllable (Table 2). The analysis cor-
rectly predicted 91% of Humboldt, 71% of Jackass, and 61% of Magel-
lanic penguin calls. Within the Spheniscus penguins, there was consid-

Thumser et al. • SPHENISCUS VOCAL COMPARISON

FACTOR(I)

Fig. 1. (A) A scatterpiot of the similarity between the bray calls of three Speniscus

species (Jacka.ss, Magellanic, and Humboldt) and an outgroup (Rockhopper penguins). (B)
A scatterpiot of the similarity between the bray calls of the three Spheniscus species. Both
scatterplots have ellipses around 50% of the data points for Humboldt (dark star), Jacka.ss
(open .square), Magellanic (dark circle), and Rockhopper (cross) penguins.

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Table 2

Discriminant Function 2 Showing the Correlations between Conditional Dependent
Variables and Dependent Canonical Factors in Three Spheniscus Species

Factor

Variable

Number of syllables

0.711

-0.020

Total duration

0.311

0.319

Inter-syllable interval

-0.136

-0.274

Duration of longest syllable

-0.225

0.735

Maximum frequency

0.441

-0.065

Minimum frequency

-0.257

0.074

Main frequency

-0.041

0.423

Chi-square

287.22

60.23

<0.001

Correlation

0.804

0.491

erable overlap in the vocalizations of Magellanic and Jackass penguins
(Fig. IB). In fact, 37% of the Magellanic penguin calls were incorrectly
classified as Jackass penguin calls, and 12% of Jackass penguin calls were
incorrectly classified as Magellanic penguin calls. By contrast, there was
less similarity in the bray call parameters of Humboldt and Jackass pen-
guins (Fig. IB). Only 6% of the Humboldt penguin calls were incorrectly
classified as Jackass penguin calls and 17% of the Jackass penguin calls
were incorrectly classified as Humboldt penguin calls. Humboldt and
Magellanic penguins were the least similar in their calls (Fig. IB). In both
species 3% of their calls were incorrectly classified as the other species.
These results were supported by the cluster analysis shown in Table 3
and Figure 2. The distance matrix and tree show the Magellanic and
Jackass penguins clustering closely together, the Humboldt penguins more
distant, and the Rockhopper penguins the most distant.

Table 3

Pearson’s Distance Matrix Showing the Distance between Four Penguin Species
Based on Seven Variables of the Bray Call

Species

1 Humboldt Penguin

2 Jackass Penguin

1.304

3 Magellanic Penguin

1.436

0.260

4 Rockhopper Penguin

1.462

1.664

1.509

Thiimser et al. • SPHENISCUS VOCAL COMPARISON

Magellanic

Jackass

Humboldt

Rockhopper

I 1 1 1

0.00 0.50 1.00 1.50

DISTANCE

Magellanic

Jackass

Humboldt

Rockhopper

King

-+-

0.00

0.20 0.40

DISTANCE

0.60

Fig. 2. (A) A tree using the single-linkage method based on Pearson’s distances of

parameters of the bray call. (B) UPGMA tree based on modified Rogers distance of allozyme
data (Thumser and Karron 1994).

DISCUSSION

Overall, the Spheniscus penguins have retained a complement of calls
that are similar in structure and function (Thumser 1993). The bray call
is used to establish a territory and to advertise availability for pairing.
The bird stands with its head pointing up and calls while slowly flapping
its wings. This was the only call which showed sufficient species-level
variation for phenetic analysis.

The analyses of selected vocal parameters of the bray call clearly dis-
tinguish Humboldt from both Jackass and Magellanic penguins. However,
discriminant and cluster analyses often could not distinguish between the
Magellanic and Jackass penguin calls. This may reflect the evolutionary
relationships among the species or may have resulted from other factors.
Since Humboldt and Magellanic penguins occur sympatrically in South
America, another possible explanation for the differences in their breeding
calls is character displacement. By contrast, the similarity of South Amer-
ican Magellanic and African Jackass penguin calls is unlikely to result
from convergence.

In order to determine whether character displacement has occurred it
is necessary to know which vocal characters are ancestral. It is difficult
to root a tree based on vocalizations and determine the most ancestral
species because vocalizations can be subject to strong selection. However,
a comparison of allozyme variation enhances these results because protein
markers are subject to weaker and different selective forces than those

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

influencing behavioral traits. Trees based on allelic characters are also
more easily rooted using outgroup taxa.

The allozyme data were consistent in two different studies (Grant et
al. 1994, Thumser and Karron 1994). Grant et al. (1994) based their
analysis on 15 polymorphic loci in 75 captive (Humboldt, Magellanic)
and wild (Jackass, Rockhopper, and Macaroni [Eudyptes chrysolophus])
penguins. Thumser and Karron (1994) studied nine polymorphic loci in
165 captive (Jackass, Humboldt, Rockhopper, King [Aptenodytes pata-
gonicus]) and wild (Humboldt, Magellanic) penguins. In both studies.
Jackass and Magellanic penguins were very closely related and Humboldt
penguins clearly formed a distinct species. There is a striking similarity
of phenetic trees based on the allozyme data and the vocal analysis pre-
sented here (Fig. 2). Cladistic analysis of the allozyme data confirmed
that Spheniscus penguins form a monophyletic group (Grant et al. 1994,
Thumser and Karron 1994). These findings suggest that differences in
vocalizations between the Humboldt and Magellanic penguins are not due
to character displacement, but rather reflect gradual genetic divergence of
separate evolutionary lineages. Although the Humboldt and Magellanic
penguins occur sympatrically, they have lower genetic identities and
greater vocal differences than the more closely related Jackass and Mag-
ellanic penguins.

ACKNOWLEDGMENTS

This research was partially funded by grants from the Ruth Walker Scholarship Lund, the
Univ. of Wisconsin— Milwaukee Graduate School, and the Institute of Museum Services (IC-
10197-91). We thank the following zoos for allowing access and providing assistance in
recording penguin vocalizations; Brookfield, Cincinnati, Henry Villas Park, Knoxville, Mil-
waukee County, Racine, and St. Louis. We are also grateful to Jim Klinesteker at the John
Ball Zoo for providing us with recordings of a Magellanic penguin population. P. D. Boers-
ma, G. Miller, and an anonymous reviewer provided several helpful suggestions on an earlier
version of the manuscript.

LITERATURE CITED

Boersma, P. E. 1974. The Galapagos penguin; a study of adaptations for life in an unpre-
dictable environment. Ph.D. diss. Ohio State Univ., Columbus, Ohio.

. 1976. An ecological and behavioral study of the Galapagos penguin. Living Bird

15;43-93.

Clancey, P. a. 1966. The penguins Spheniscus clemersus Linnaeus and Spheniscus niagel-
lanicus Forster. Ostrich 37;237.

Eggleton, P. and W. R. Siegfried. 1979. Displays of the Jackass penguin. Ostrich 50; 139-
167.

Grant, W. S., D. C. Duffy, and R. W. Leslie. 1994. Allozyme phylogeny of Spheniscus
penguins. Auk 1 I 1;7 16-720.

JouvENTiN, P. 1982. Visual and vocal signals in penguins, their evolution and adaptive
characters. Adv. Ethology 24.

Thumser et al. • SPHENISCUS VOCAL COMPARISON

Miller, E. H., W. W. H. Gunn, and B. N. Veprintser. 1988. Breeding vocalizations of
Baird’s Sandpiper Calidris bairdii and related species, with remarks on phylogeny and
adaptation. Ornis Scand. 19:257-267.

O’Hare, R. J. 1989. Systematics and the study of natural history, with an estimate of the
phylogeny of living penguins (Aves: Spheniscidae). Ph.D. diss. Harvard Univ., Boston,
Massachusetts.

Payne, R. B. 1986. Bird songs and avian systematics. Pp. 87—116 in Current ornithology.
Vol. 3 (R. E Johnston, ed.). Plenum Press, New York, New York.

Pettingill, O. S. Jr. 1960. Creche behavior and individual recognition in a colony of
Rockhopper penguins. Wilson Bull. 72:209—221.

PiEROTTi, R. 1987. Isolating mechanisms in seabirds. Evolution 41:559-570.

SCOLARO, J. A. 1987. A model life table for Magellanic penguins {Spheniscus magellanicus)
at Punta Tombo, Argentina. J. Eield Ornithol. 58:432-441.

Sibley, C. G. and B. L. Monroe, Jr. 1990. Distribution and taxonomy of birds of the
world. Yale Univ. Press, New Haven, Connecticut.

Sigma Scan. 1988. Scientific Measurement Program, Version 3.90. Jandel Scientific, Corte
Madera, California.

Simpson, G. G. 1976. Penguins past and present, here and there. Yale Univ. Press, New
Haven, Connecticut.

Spurr, E. B. 1975. Communication in Adelie Penguins. Pp. 449-501 in The biology of
penguins (B. Stonehouse, ed.). Macmillan, New York, New York.

Stonehouse, B. 1960. The King Penguin (Aptenodytes patagonicus) of South Georgia.
Talk. Isl. Dep. Surv. Sci. Rep. 23.

Thumser, N. N. 1993. Phylogenetic relationships among Spheniscus penguins based on the
analysis of vocal and allozyme data. Ph.D. diss. Univ. of Wisconsin-Milwaukee, Mil-
waukee, Wisconsin.

AND J. D. Karron. 1994. Patterns of genetic polymorphism in five species of

penguins. Auk 111:1018-1022.

Wilkinson, L. 1990. SYSTAT: the system for statistics. SYSTAT, Inc., Evanston, Illinois.

Wilson Bull., 108(1), 1996, pp. 80-93

THE BREEDING BIOLOGY OF THE WILLOW TIT IN
NORTHEASTERN SIBERIA

Vladimir V. Pravosudov and Elena V. Pravosudova

Abstract. We studied the breeding biology of the Willow Tit {Parus montanus) during

19g7_1990 in the Magadan region of northeastern Siberia. Clutch size and number of fledg-
lings averaged 7.5 and 6.5, respectively, and both were correlated negatively with the date
of the first egg. Nestling growth rate was correlated positively with the date of the first egg,
but was not related to brood size. Body mass at fledging was related negatively to brood
size. Males fed nestlings significantly more often than did females, while females spent
more time attending nestlings. The number of parental visits per young did not change
significantly with brood size. During the first 13 days of the nestling period, female feeding
rate per young was positively related to brood size while for males this relationship was
negative. The nestling diet consisted mostly of Lepidoptera larvae, Arachnoidea, and Dip-
tera. Received 20 April, 1995, accepted 21 Sept. 1995.

Most studies of the Willow Tit {Parus montanus), a small hole-nesting
Palearctic passerine, have focused on the species’ non-breeding biology
in northern Europe (see Ekman 1989, Matthysen 1990). Eewer studies
have been done on its breeding biology (Orell 1983, Orell and Ojanen
1983, Pravosudov 1987, Orell and Koivula 1988, see review by Cramp
and Perrins 1993). Here, we report details of breeding biology of the
Willow tit in northeastern Siberia and compare it with breeding biology
of other parids. Our study area is close to the easternmost part of this
species’ range.

METHODS

We collected data between 1987 and 1990 in the .southern part of the Magadan legion of
northeastern Siberia (60°N, 150°E). The habitat was comprised mainly of larch {Lati.x ca-
jandery), poplar (Populus suoveolens), and chosenia {Chosenia arbutifolia). A detailed de-
scription of the study area has been published previously (Pravosudov 1993a, b). Most of
the birds were fitted with unique combinations of color bands. Nests were opened with a
knife, after which a patch of bark was used to cover the opening. Nestlings in 13 nests were
weighed to the neare.st 0.1 g and the fifth primary was measured to the nearest millimeter.
Using nonlinear regression, we found that the logistic equation was the most suitable for
describing rate of body mass increase, and we used the growth constant K to estimate
nestling growth rate. As another index of nestling growth, we used the growth of the fifth
primary. The rate of increase in length of this feather is very close to linear, so we used the
slope of the linear regression as an estimate of the growth rate. The rates at which nestlings
were fed were studied by repeatedly observing eight nests for 1-h intervals over the whole
nestling period. All observations were made from 35^0 m, and the parents did not appear

Institute of the Biological Problems of the North, Academy ol Sciences of Russia, Far East Division, K.
Marx pr. 24, Magadan, Russia. (Present address; Behavioral Ecology Group, Department ot Zoology,
The Ohio State University, 1735 Neil Avenue, Columbus, Ohio, 43210-1293.)

Pravosudov and Pravosudova * WILLOW TITS

to be disturbed by our presence. To collect food brought to each nestling by its parents, we
used a thread tied loosely around each nestling’s neck. After the parents made a number of
feeding visits corresponding to the number of young in a nest, we collected the food loads
from each nestling’s throat with forceps and stored each load in a separate vial. Parents
resumed feeding as soon as collared nestlings were put back in the nest. This method did
not appear to harm the young tits. In 1987, we determined only the frequency of different
prey items in the diet. In 1989 and 1990, we also weighed to the nearest mg each food load
taken from a nestling and each individual item such loads contained. Over all years, we
collected 296 food items contained in 108 individual loads from 11 broods, 185 of which
were weighed. The number of food loads collected per nest was 5—23 loads. The food loads
were collected from nestlings 6—12 days old, with most collected during 7—11 days of
nestling age.

Multi-way general linear models (GLM) and multiple and simple linear regression anal-
yses were used for the majority of tests. General linear model (GLM) is a multivariate
analysis that is used to perform analysis of variances (ANOVA) with balanced and unbal-
anced designs, analysis of covariance (ANCOVA), and regression (Neter et al. 1990, Anon-
ymous 1991). All analyses were performed using MINITAB routines (Anonymous 1991).
All tests of the slopes in regression analyses and for covariates in the GLMs were two-
tailed. For statistical analyses of parental feeding rates, we used two methods (1) we ana-
lyzed averages per nest for the whole nestling period (16 days), and (2) we used all obser-
vations (147 1-h periods) made repetitively at the eight nests in a GLM where each nest
served as a factor. All models dealing with parental feeding rate consisted of a nest as a
factor and nestling age and brood size as covariates. Analyses of the length of time that
parents spent in the nest during one visit while feeding nestlings was done on three nests
only, and all 579 such observations were used in a GLM where each nest served as a factor
and brood size and nestling age served as covariates. For statistical analyses of variance in
food loads and main prey type in the diet among different years and different broods, we
used average frequency of every prey type in a load per nest in ANOVA. Thus, the brood
was the primary sampling unit. To analyze variation in the diet among different broods, we
used average frequency of every prey type in a load and the load was the primary sampling
unit. The variation among broods was tested separately for each year.

RESULTS

Willow Tits almost always excavated their own cavities. Only two of
22 nests were in pre-existing natural holes, and in both cases, the birds
altered the original hole by shaping the cavity. The average start of egg-
laying differed significantly among years, occurring a week later during
1990 than during the two preceding years (ANOVA, ^2,8 = 5.7, P =
0.012; Table 1). Mean clutch size was 7.5 eggs and varied significantly
among years. Birds that started laying eggs later had smaller clutches
(ANCOVA, effect of year, F, ,7 = 10.1, P = 0.006; date of the first egg,
slope = -0.15, P = 0.006; Table 1). The number of young that fledged
was 6.5 and did not vary significantly among years, but pairs that started
laying eggs later had significantly fewer fledglings (ANCOVA, effect of
year, F, ,6 = 2.4, P = 0.124; date of the first egg, slope = -0.19, P =
0.03; Table 1). Reproductive success as fledglings per egg averaged 0.85
and was nearly always higher than 0.7 (Table 1). The nestling body mass

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Table

Breeding Characteristics of Willow Tits

IN THE Magadan Region of Siberia*"

Parameter

1987

1989

1990

Total

Range

No. of nests

Mean date of first egg

30 May

6 June

1 June

25 May-7 June

1.6 days

1.9 days

1.7 days

5.6 days

Clutch size

7.1

8.0

7.7

7.5

6-10

0.4

0.5

0.4

1.1

No. of fledglings

6.4

6.2

6.8

6.5

3-9

0.6

0.7

1.6

Reproductive success**

0.89

0.75

0.89

0.85

0.42-1.00

5.2

6.1

15.5

K constant

0.376

0.372

0.411

0.385

0.331-0.436

0.017

0.015

0.016

0.035

Primary growth rate

3.28

3.44

3.45

3.39

3.02-3.60

(mm/day)

0.07

0.06

0.07

0.16

Body mass of young

11.91

12.49

12.54

12.33

11.34-13.16

on day 14 posthatch

0.43

0.69

0.27

0.56

•‘Within each cell of the four rows on the left, the upper number is the mean and the lower number is the standard
deviation.

Reproductive success = fledglings/eggs.

growth constant (K) did not vary significantly among years or with brood
size (ANOVA, P > 0.2). The nestlings in the later nests, however, grew
significantly faster (regression on the date of the first egg, b = +0.005,
P - 0.011, constant K = 0.213 + 0.005 date; Fig. 1). Body mass of
nestlings on day 14 posthatch did not vary significantly among years
(Table 1) or with date of first egg (ANOVA, P > 0.2). However, day- 14
body mass appeared to be related to brood size; nestlings in smaller
broods were heavier (regression on the brood size, b = —0.163, P =
0.064, N = 13, mass = 13.4 — 0.163 brood size). Body mass of fledglings
was not significantly related to their growth rate (regression, P = 0.80).
The growth rate of the fifth primary varied similarly with the growth
constant, K. Feather growth was not significantly affected by either year
or brood size (ANOVA, P > 0.2), but the nestlings in later broods tend
to grow their fifth primary faster (regression on the date of the first egg,
b = 0.017, F = 0.082, N = 13, rate = 2.83 + 0.017 date).

Males made significantly more feeding trips to nest than did females
(paired t-test, t — 3.56, N = 8 nests, P = 0.009, Fig. 2). Females increased
the number of feeding trips per brood as brood size increased (Regression,
b = 0.82, P = 0.046, N = 8 broods, number of trips = 1.18 + 0.82
brood size), while total number of visits and number of visits by males
per nest were not significantly affected by brood size (F > 0.2, calculated

Pravosudov and Pravosudova • WILLOW TITS

CZJ

0.45

0.44

0.43

0.42

0.41

0.40

0.39

0.38

0.37

0.36

0.35

0.34

0.33

0.32

0.31

0.30

I I I I L

25 30 4 9 14

MAY JUNE

DATE OF THE FIRST EGG

Fig. 1 . Relationship between body mass growth rate constant, K, of Willow Tits and
the date of the first egg.

on averages per nest for the entire nestling period). GLM analysis using
all feeding observations and brood as a factor during the entire nestling
period showed no significant relationships between the number of visits
per nest (total, male and female separately) and brood size {P > 0.2).
Both total number of visits per nest (slope = 3.12, P < 0.01) and number
of visits by females (slope = 3.95, P < 0.01) significantly increased as
nestlings grew older, although male feeding rate was affected by nestling
age only suggestively (slope = 1.56, P = 0.095). The number of feeding
trips per young (total, male and female separately) was also not signifi-
cantly affected by brood size {P > 0.4, calculated both for nest averages
for all nesting period and with a nest as a factor for all feeding obser-
vations). Both total number of feeding trips (GLM with a brood as a

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Eig. 2. Number of feeding visits of Willow Tits per nestling per hour as related to
nestling age. Open circles and squares represent feeding rates of males and females, re-
spectively. Triangles represent rates for males and females combined. Vertical lines represent
± one SD of the means.

factor, slope = 2.64, P < 0.01) and number of visits by female per young
(slope = 3.30, P < 0.01) increased significantly as nestlings grew older,
while male feeding rate was not significantly affected by nestling age
(slope = 1.29, P = 0.22). Since some young left their nest on the 14th
day posthatch and nestling body mass (Fig. 3) and parental feeding rate
(Fig. 3) started slowing down by the 13th day posthatch, we analyzed the
frequency of parental feeding visits separately for the first 13 days of
nestling age (N = 116 1-h periods for 8 nests). This separate analysis of
the first 13 days may be very important since the young reach their fledg-
ing body mass and are able to fledge by that time, so the heaviest pressure
on parents should fall in this period. The results were strikingly different

Prcivosudov and Pravosndova • WILLOW TITS

Fig. 3. Nestling body mass of Willow Tits as related to nestling age.

from those of the whole nesting period (Table 2). Both total number of
feeding trips and number of feeding trips by female per brood and per
young increased significantly as nestlings grew older. Male feeding rate,
though, did not change significantly with nestling age (Table 2). Both
males and females increased number of feedings per nest as brood size
increased. Feeding frequency per young showed a dramatic difference
between sexes (Fig. 4). Females made more feeding trips per young in
larger broods, while males made significantly fewer of them as brood size
increased (Table 2, Fig. 4).

Females, but not males, spent more time attending nests containing
fewer young during the entire nesting period (GLM with nest and brood
size as factors, ^2,579 = 8.27, P < 0.001). Both male and female decreased
their attendance time as nestlings grew older (GLM with a nest as a factor,
P < 0.01, Fig. 5).

Arachnoidea, Lepidoptera, and Diptera comprised the majority of the

THE WILSON BULLETIN • Vol. 108, No. I, March 1996

Table 2

Relationships between Parental Leeding Rate and Nestling Age (Day 1-13) and

Brood Size in the Willow Tit

No. of visils/h

Age

Brood

size

Slope

Male per brood

0.16

0.13

1.10

0.00

Lemale per brood

0.44

0.00

1.37

0.00

Total per brood

0.59

0.00

2.49

0.00

Male per young

0.03

0.11

-0.12

0.01

Lemale per young

0.06

0.00

0.08

0.01

Total per young

0.09

0.00

-0.03

0.58

nestlings’ diet (79%, Tables 3 and 4). The number of prey items contained
in each individual food load brought to nestlings averaged 3.0 and did
not vary significantly among years (Fj g = 1.76, P = 0.23, Table 5) or
different broods (no two pairs were different at F = 0.05, Tukey’s test).
Average mass of a load taken during 1989-1990 also was not statistically
different (r-test, t = 1.74, P = 0.22, N = 7) between these years (Table
3), although there were significant differences among broods during both
years (in 1989, one nest was different from one out of 3 broods and in
1990, one nest was different from the rest, all differences at F = 0.05,
Tukey’s test). The mass of a food load was positively and significantly
related to the number of prey items in a load when differences among
years and broods were accounted for (ANCOVA with year and brood as
factors and the number of items in a load as a covariate, individual load
is the primary sampling unit; slope = 15.64, t = 5.49, F < 0.001). Even
though all major prey items showed a great deal of variation there were
no significant difference in average frequency of any major prey type per
load either among years or among broods (no pairs compared were dif-
ferent at F = 0.05, Tukey’s test).

DISCUSSION

The breeding biology of the Willow Tit in northeastern Siberia and
western Europe appears to be very similar (Orell and Ojanen 1983). Av-
erage clutch size (7.62) and number of fledglings (6.19) in Einland (Orell
and Ojanen 1983) are nearly identical to those in Siberia (7.5 eggs and
6.5 fledglings). The trend toward reduced clutch size with later breeding
seems to be general not only for Willow Tits but for many other bird
species as well (Orell and Ojanen 1983). Nestling growth rate (K) of
Siberian Willow Tits (0.331-0.436) is very similar to that of Willow Tits

Pravosudov and Pravosudova • WILLOW TITS

CL.

2.2

2.0

Fig. 4. Feeding rates by male and female Willow tits per brood (A) and per nestling
(B) as related to brood size. Feeding rates are taken as averages for the first 13 days of
nestling age, and lines represent standard deviations of the means. Males are represented by
circles and females by squares.

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Eig. 5. Eength of time spent in the nest cavity during one visit by male and female
Willow Tits as related to nestling age. Vertical lines represent standard deviations of the
means. Males are shown by circles and females by squares.

in other studies (0.380-0.406; Foster and Godfrey 1950, Inozemtsev
1962, Song 1980, Orell and Ojanen 1983). The negative coirelation be-
tween fledgling body mass and brood size has also been demonstrated
before (Nur 1984, Orell and Koivula 1990).

It seems that birds that start breeding earlier typically produce more
fledglings. However, we found that nestling growth rate was positively
correlated with the start of breeding. Nestlings that hatched later grew
significantly faster although there was no effect of the timing of breeding
on fledgling body mass. The fledgling body mass was also unrelated to
their growth rate. This result may suggest that although late-breeding pairs
produce fewer fledglings, their young could be better off nutritionally.
Timing of breeding has long been thought of as a life-history trade-off.

Pravosudov and Pravosudova • WILLOW TITS

Table 3

Frequency of Different Food Types in the Diet of Nestling Willow Tits in Siberia

Prey taxon

1987

1989

1990

Total

Percent

Arachnoidea

16.5

Total Lepidoptera

135

45.6

Larvae

122

41.2

Imago

4.4

Total Hymenoptera

6.8

Larval Hymenoptera

0.7

Tentridinidae

0.3

Pamphilidae

0.3

Formicidae

1.0

Total Coleoptera

7.4

Larval Coleoptera

6.4

Diptera

16.9

Limoniidae

8.4

Tipulidae

7.1

Unidentified

1.3

Plecoptera

2.0

Chloroperlidae

1.0

Unidentified

1.0

Homoptera

1.0

Gastropoda

0.7

Fish bones

0.7

Total number

111

1 14

296

Birds may benefit by breeding earlier, possibly because food abundance
is higher. However, for the Willow Tits breeding in Siberia, this is prob-
ably not true since when they start breeding there is still snow cover and
food becomes abundant only later (pers. obs.). Other studies have shown
that in some resident passerine birds, including the Willow Tit, there is
another important benefit from breeding earlier; young fledged earlier
stand a better chance of recruitment into both winter flocks and into the
next season’s breeding population (Nilsson 1988, Koivula et al. 1993,
Pravosudov 1993b). Our results suggest that there may be a trade-off
between early and late breeding in the Willow Tit. If nutritional condition
of fledglings were independent of the start of breeding, birds breeding
earlier would produce young that would have a higher chance of estab-
lishing themselves into a subsequent breeding population. However, birds
breeding later may produce better nourished young, although fewer of
them. The optimal time of breeding may thus reflect a trade-off between
breeding early enough to produce early dispersers (and successful re-
cruits) and breeding late enough to produce well nourished young.

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Table 4

Mass (mg) of Different Lood Types in the Diet of Nestling Willow Tits

Prey taxon

1989

1990

Total

Percent

Arachnoidea

833

702

1535

20.5

Lepidoptera

2392

2531

4923

65.8

Larvae

2175

2531

4706

62.9

Imago

217

2.9

Hymenoptera

160

245

3.3

Pamphilidae

0.8

Lormicidae

1.1

Larvae

100

1.3

Coleoptera

104

161

2.1

Larvae

104

140

1.9

Diptera

310

119

429

5.7

Tipulidae

290

119

409

5.5

Plecoptera

1.1

Chloroperlidae

0.9

Homoptera

1.0

Gastropoda

0.4

Total

3928

3555

7483

Parental behavior in which male Willow Tits fed young more often
than females but females spent more time attending nestlings during the
whole nesting period was quite similar to that of the Mountain Chickadee
{Pams gambeli) (Grundel 1987). However, during the first 13 days of
nestling life. Willow Tits differed significantly from Mountain Chickadees
in the way the sexes responded to changes in brood size. In our study,
both males and females increased the number of feeding visits per brood
as the brood size increased. However, the number of feeding trips per
young was negatively correlated with brood size in males but positively
correlated in females. This result, and also the fact that male feeding rate

Table 5

Measurements of Nestling Leeding in the Willow Tit

Parameter

1987

1989

1990

Total

Number of loads

108

Number of broods

1 1

Number of food items/load

4.2 (2.6)“

2.2 (0.7)

2.5 (0.8)

3.0 (1.8)

Mass of a food load

—

81.1 (22.2)

127.9 (42.4)

101.1 (38.4)

" so.

Pravosudov and P ravosudova • WILLOW TITS

did not change much with nestling age, may indicate that males worked
as hard as possible and that is why their feeding rate was similar for
broods of different size and age throughout the entire nesting period. On
the other hand, females appeared to adjust their feeding rate to variation
in demand caused by different ages of young and different brood sizes.
Females, but not males, also adjusted their attendance time to brood size
while feeding nestlings.

It is known that parental feeding rate by itself is not necessarily a good
indicator of parental investment since the biomass of food brought by
parents to their young per visit may differ between males and fernales
(Grundel 1987). However, Grundel (1987) found in the Mountain Chick-
adee that differences in total volume of food per nestling in broods of
different size were due to changes in feeding frequency rather than in
prey size or load size. Therefore, the patterns of feeding frequencies by
male and female Willow Tit parents of large and small broods appear to
be a valid representation of their investment. The pattern of feeding rate
with an increase in the beginning of the nestling stage, a plateau in the
middle, and a decrease before fledging seems to be very common and
has been shown for many parids (Gibb 1950, Royama 1966, Grundel
1987). Because of the sexes’ opposite trends in feeding rate per young
with an increase in brood size, the total number of feedings per young in
Siberian Willow Tits was not significantly different in broods of different
size. This pattern has not been found in many studies (Gibb 1950, 1955;
Royama 1966, Walsh 1978, Grundel 1987), although it has been dem-
onstrated before (Pinkowski 1978). Assuming that the changes in feeding
rate associated with changes in brood size truly represent a change in
parental investment, our results are consistent with the individual adjust-
ment hypothesis (Hogstedt 1980, Pettifor et al. 1988) which assumes that
birds adjust their clutch size to their own capabilities of raising young.
The absence of any relationship between brood size and nestling growth
rate also appears to support this hypothesis. The negative relationship
between fledgling body mass and brood size seems to go against the
individual adjustment hypothesis since post-fledging survival is known to
correlate positively with fledging body mass (Nur 1984, Orell and Koivula
1990). Since young in large broods tend to be lighter compared with
young from smaller broods, one can assume that those from larger broods
should have lower survival. However, experimental results from studies
of Willow Tits in Finland did not fully support this assumption, suggest-
ing that any relationship among brood size, nestling weight, and juvenile
survivorship can be complicated by environmental variability (Orell and
Koivula 1990). Also, from all the relationships described above, we can
assume that lighter young fledge earlier (earlier breeding start results in

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

larger brood size which results in lighter body mass of fledglings) and,
if they survive, would have a greater chance of breeding.

ACKNOWLEDGMENTS

We thank Elena Zimireva for her assistance in the field. T. C. Grubb, Jr., R. A. Mauck,
and T. A. Waite, O. Hogstad, and C. R. Blem provided valuable comments on an earlier
draft of the manuscript.

LITERATURE CITED

Anonymous. 1991. Minitab reference manual: release 8. Minitab, State College, Pennsyl-
vania.

Cramp, S. and C. M. Perrins. 1993. The birds of the western Palearctic. Volume VII.
Flycatchers to Shrikes. Oxford Univ. Press, Oxford, England.

Ekman, J. 1989. Ecology of non-breeding social systems of Parus. Wilson Bull. 101:263—
288.

Foster, J. and C. Godfrey. 1950. A study of the British Willow Tit. Brit. Birds 43:351-
361.

Gibb, J. A. 1950. The breeding biology of the Great and Blue titmice. Ibis 92:507—539.

. 1955. Feeding rates of Great Tits. Br. Birds 48:49-58.

Grundel, R. 1987. Determinants of nestling feeding rates and parental investment in the
Mountain Chickadee. Condor 89:319-328.

Hogstedt, G. 1980. Evolution of clutch size in birds: adaptive variation in relation to
territory quality. Science 210:1148-1150.

Inozemtsev, A. A. 1962. Notes on the ecology of tits in the Moscow region. Pp. 169-199
in Materialy po faune i ekologii zhivotnykh. (S. P. Naumova, ed.). (in Russian).

Koivula, K., K. Lahti, M. Orell, and S. Rytkonen. 1993. Prior residency as a key
determinant of social dominance in the Willow Tit {Parus montaniis). Behav. Ecol.
Sociobiol. 33:283-287.

Matthysen, E. 1990. Nonbreeding social organization in Parus. Curr. Ornithol. 7:209-249.

Neter, j., W. Wasserman, and M. H. Kutner. 1990. Applied linear, statistical models.
IRWIN Press, Homewood, Illinois.

Nilsson, J.-A. 1988. Causes and consequences of dispersal in Marsh Tits: time as a fitness
factor in establishment. Ph.D. diss.. University of Lund, Lund, Sweden.

Nur, N. 1984. The consequences of brood size for breeding Blue Tits. II. Nestling weight,
offspring survival and optimal brood size. J. Anim. Ecol. 53:497-517.

Orell, M. 1983. Nestling growth in the Great Tit Parus major and the Willow Tit P.
montanus. Ornis Fenn. 60:65-82.

AND M. Ojanen. 1983. Breeding biology and population dynamics of the Willow
Tit Parus montanus. Ann. Zool. Fenn. 20:99-1 14.

AND K. Koivula. 1988. Cost of reproduction: parental survival and production of
recruits in the Willow Tit Parus montanus. Oecologia 77:423^32.

AND . 1 990. Effects of brood size manipulations on adult and juvenile sur-
vival and future fecundity in the Willow Tit, Parus montanus. Pp. 297—306 in Popu-
lation biology of passerine birds, an integrated approach (J. Blondel, A. Gosler, J.-D.
Lcbreton, and R. McCleery, cds.), NATO ASI Series, vol. G-24. Springer- Verlag, Hei-
delberg, Germany.

Peitifor, R. a., C. M. Perrins, and R. H. McCleery. 1988. Individual optimization of
clutch size in Great Tits. Nature 336:160-162.

Pruvosudov and Pravosudova • WILLOW TITS

PiNKOWSKi, B. C. 1978. Feeding of nestling and fledgling Eastern Bluebirds. Wilson Bull.
90:84-98.

Pravosudov, V. V. 1987. Ecology of two closely related species of tits in the northwestern
part of the USSR. Ornitologia (Moscow) 22:68-75 (in Russian).

. 1993a. Breeding biology of the Eurasian Nuthatch in northeastern Siberia. Wilson

Bull. 105:475-482.

. 1993b. Social organization of the Nuthatch Sitta europaea asiatica. Ornis Scand.

24:290-296.

Royama, T. 1966. Factors governing feeding rate, food requirement and brood size of
nestling Great Tits Parus major. Ibis 108:313—347.

Song, Y. 1980. Studies on the breeding ecology and feeding habits of Willow tits. Acta
Zool. Sinica 26:370-377 (in Chinese with English summary).

Walsh, FI. 1978. Food of nestling Purple Martins. Wilson Bull. 90:248-260.

Wilson Bull., 108(1), 1996, pp. 94-103

CENSUSING WINTERING POPULATIONS OF
SWAINSON’S WARBLERS: SURVEYS IN THE
BLUE MOUNTAINS OF JAMAICA

Gary R. Graves

Abstract. — Census methods developed for breeding populations of Nearctic-Neotropic
migrant passerines are largely ineffective for determining the distribution and abundance of
Swainson’s Warbler {Limnothlypis swainsonii) on its wintering grounds in the Caribbean
basin. Using playback of tape-recorded call notes interspersed with advertising songs, I
found the warbler to be widespread and relatively common in montane forests of the Blue
Mountains of Jamaica. Detection rates with playback varied from 17.8 to 29.2 warblers/10
h along five census transects. Census efficiency was increased by an estimated factor of five
to 10 times with the use of tape playback. Received 23 March 1995, accepted 20 Aug. 1995.

For conservationists, the apparent decline of some Nearctic-Neotropic
migratory birds is a serious concern (Terborgh 1989, Robbins et al. 1989)
that demands both rigorous management and quantitative monitoring of
breeding and wintering populations. Effective census techniques have
been developed for many songbird species (Passeriformes) during the
breeding season (Ralph and Scott 1981, Robbins et al. 1986). However,
songbirds rarely sing on their wintering grounds and during fall migration,
are feathered in duller basic plumages, and hence, are more difficult to
detect and identify than during the breeding season. Yet most studies of
migrants in the Neotropics employ censusing techniques designed for
breeding birds of higher latitudes, with little or no attempt to compensate
for the decreased detectability of wintering populations. As this report
will communicate, standard point count and transect censusing methods
can be relatively ineffective in determining the true abundance of secre-
tive migratory species that winter in dense tropical habitats.

The breeding and wintering biologies of Swainson’s Warbler {Lim-
nothlypis swainsonii) are among the most poorly known of the migratory
wood warblers (cf Meanley 1971, Eddleman 1978, Graves 1992, Brown
and Dixon 1994, Thomas 1994). Wintering birds have been reported in
the Caribbean basin from the Bahamas, Cuba and the Isle of Pines, Ja-
maica, Cayman Islands, and the Yucatan Peninsula, casually east to Puerto
Rico and the Virgin Islands, and from southern Veracruz south to Hon-
duras (Meanley 1971, AOU 1983, Pashley 1988, Winker et al. 1992).
Most wintering records are anecdotal (e.g., Eaton 1953), and, at present,
wintering sites where more than one or two individuals can be consis-

Dept. of Vertebrate Zoology, National Museum of Natural History, Smithsonian Institution, Washington,
D.C. 20560.

Graves • CENSUSING SWAINSON’S WARBLERS

tently encountered per day are known only from certain regions of Cuba
(G. Wallace, pers. comm.; Kirkconnell et al., unpubl. ms.). This paper
documents the utility of playback techniques for censusing wintering pop-
ulations of Swainson’s Warbler and reports the discovery of what appears
to be a major wintering area for the species in the Blue Mountains of
Jamaica.

METHODS

Study area. — I surveyed Swainson’s Warbler populations along the same trails and roads
at Hardwar Gap that were studied by the Lacks, from 2 to 10 February 1995. Hardwar Gap,
at the divide forming the boundary between St. Andrew and Portland Parishes in the Port
Royal Mountains (an outlier of the Blue Mountains), is accessible along a hard-surfaced
road (B 1 ) from Kingston to Buff Bay on the northern coast. Hardwar Gap is one of the
most frequent destinations in Jamaica of resident and visiting ornithologists and bird watch-
ers. Both Hardwar Gap and the adjacent Hollywell Park now lie within the borders of the
Blue and John Crow Mountains National Park (Muchoney et al. 1994).

Three census transects were established along trails (“forest” transects in Table 1) that
pass through a peninsula of upper montane rain forest connected to a much larger block of
forest eastward in the national park (see Asprey and Robbins 1953 and Muchoney et al.
1994 for descriptions of habitat). The Hardwar Gap region was extensively affected by
Hurricane Gilbert in 1988 (Wunderle et al. 1992), and patches of early successional vege-
tation were present along all three forest transects, but there was little evidence of cutting
or other human disturbance away from roadsides. Transect terrain was extremely steep.
Slopes varied from 15° to 70° and were estimated to average over 40°. Significant portions
of forest trail No. 3 (Table 1) were at the summit of a knife-like ridge.

The fourth and fifth census routes followed the main road (Bl) from a point just below
Hardwar Gap to the village of Section, Portland Parish. This road forms the boundary
between relatively undisturbed montane forest (upslope) and a patchwork of lightly modified
montane forest, coffee plantations, residential gardens, and agricultural scrub (downslope)
in the 1 km buffer zone of the national park. The upper and lower sections of the road,
although contiguous, were treated as separate census routes (Table 1).

The length of “forest” transects was estimated by averaging the number of steps counted
on three downslope trips (at 0.76 m/step). The road transects were measured with a vehicle
odometer. Elevations were estimated by multiple readings of a calibrated altimeter.

Census methods. — I used techniques developed for monitoring breeding populations of
Swainson’s Warblers in the southeastern United States (Graves, unpubl.). Loop cassette tapes
(20 s and 60 s) were prepared from recordings (made a with Marantz PMD430 cassette
recorder and a Sennheiser ME 80 directional microphone in Virginia and Louisiana) of
Swainson’s Warbler call notes interspersed with a primary song every 15 s (one song on 20
s loop). Advertising songs were included because they are more easily heard above the low
frequency noise caused by ru.stling leaves and dripping water in the forest understory. The
opposite proportion of songs and calls are used to census breeding populations.

Recordings were broadcast from a dual-speaker “boom box” (Sanyo). Power output was
variable due to battery use and atmospheric conditions, but the audio output was adjusted
before every census run so that call notes were audible to me at a distance of 60 m and
songs at a somewhat greater distance (unobstructed by vegetation). The direction of the
speakers was rotated every 5 to 10 s until a probable response was detected. During clear
weather at Hardwar Gap, Swainson’s Warblers responded to tape broadcasts from distances

THE WILSON BULLETIN • Vol. 108, No. 1, March 1996

Table 1

Censuses of Swainson’s Warblers at Hardwar Gap, Blue Mountains, Jamaica

Census transect

Census

run

Starting

time

Census

duration

(m)

Weather'

Number

recorded

Cumu-

lative

number

Lorest trail